January 2017
This publication was prepared independently by Jordan Robinson, Daniel Sabet, Meredith Feenstra, Stephanie Stahlberg, 
and Olga Rostapshova through Social Impact, Inc. It was produced at the request of the United States Agency for 
International Development as part of the Democracy, Human Rights and Governance Learning, Evaluation, and Research 
activity.
Impact Evaluation of the “Increasing Services for 
Survivors of Sexual Assault in South Africa” Program
ENDLINE IMPACT EVALUATION REPORT 
Impact Evaluation of the “Increasing Services 
for Survivors of Sexual Assault in South Africa” 
Program 
ENDLINE IMPACT EVALUATION REPORT
January 2017
AID-OAA-M-13-00011
DISCLAIMER
The authors’ views expressed in this publication do not necessarily reflect the views of the United States Agency for 
International Development or the United States Government.
CONTENTS
Acronyms.................................................................................................................................................................................................. ii
Tables........................................................................................................................................................................................................ iii
Figures....................................................................................................................................................................................................... iv
Executive Summary................................................................................................................................................................................. v
Introduction......................................................................................................................................................................................... v
Evaluation Questions and Design .................................................................................................................................................. v
Findings and Conclusions.............................................................................................................................................................. viii
TCC capacity concerns............................................................................................................................................................ viii
No impact of community dialogues and service provider training on TCC utilization........................................... viii
No impact of community dialogues on community knowledge and attitudes.............................................................ix
Positive impact of training on service providers’ knowledge and attitudes..................................................................ix
Recommendations............................................................................................................................................................................. x
Introduction.............................................................................................................................................................................................. 1
1.1 Evaluation purpose ..................................................................................................................................................................... 1
1.2 Project background .................................................................................................................................................................... 1
Evaluation Design .................................................................................................................................................................................... 4
2.1 Evaluation questions and hypotheses..................................................................................................................................... 4
2.2 Sampling, randomization, and power calculations.............................................................................................................. 6
2.2.1 Precinct level ....................................................................................................................................................................... 6
2.2.2 Individual level ..................................................................................................................................................................... 8
2.3 Data sources and data limitations........................................................................................................................................... 9
2.3.1 Police records...................................................................................................................................................................... 9
2.3.2 TCC records and supplemental intake form............................................................................................................... 9
2.3.3 Women’s survey...............................................................................................................................................................11
2.3.4 Service provider survey ..................................................................................................................................................11
2.3.5 Qualitative TCC and NGO data ..................................................................................................................................11
2.3.6 Training, piloting, and data collection..........................................................................................................................12
Findings....................................................................................................................................................................................................13
3.1 TCCs............................................................................................................................................................................................13
3.2. TCC capacity.............................................................................................................................................................................14
3.2.1 Days and hours of operation.........................................................................................................................................14
3.2.2 After-hours services........................................................................................................................................................14
3.2.3 Staffing .................................................................................................................................................................................15
3.2.4 Location ..............................................................................................................................................................................16
3.2.5 Ability to meet survivor needs......................................................................................................................................16
3.2.6 Communication and collaboration with stakeholders............................................................................................17
3.2.7 Space constraints..............................................................................................................................................................17
3.2.8 Facility appearance and supplies...................................................................................................................................18
3.2.9 Medical supplies, clothing, and food ............................................................................................................................19
3.2.10 Transportation................................................................................................................................................................20
3.2.11 Funders and supporters................................................................................................................................................21
3.3 TCC utilization ..........................................................................................................................................................................21
3.3.1 Survivor demographics....................................................................................................................................................21
3.3.2 Referrals to TCCs............................................................................................................................................................22
3.3.3 Barriers to accessing TCCs...........................................................................................................................................23
3.3.4 NGOs as service providers...........................................................................................................................................24
3.4 Impact of interventions on TCC utilization .......................................................................................................................26
3.4.1 Description of supplemental intake data submitted by TCCs..............................................................................26
3.4.2 Balance statistics at baseline ..........................................................................................................................................27
3.4.3 Impact of community dialogues on TCC utilization................................................................................................27
3.4.4 Impact of service provider training on TCC utilization..........................................................................................30
3.5 Impact of community dialogues on community knowledge and attitudes ................................................................. 32
3.5.1 Balance statistics...............................................................................................................................................................32
3.5.2 Knowledge of TCCs and other services in the community to victims of sexual assault...............................34
3.5.3 Perceptions of sexual assault and sexual assault reporting ...................................................................................42
3.5.4 Attitudes toward gender................................................................................................................................................46
3.6 Impact of training on service providers’ knowledge and attitudes ..............................................................................50
3.6.1 Knowledge of TCCs........................................................................................................................................................51
3.6.2 Recent services provided to SGBV survivors...........................................................................................................53
3.6.3 Attitudes about SGBV.....................................................................................................................................................57
3.6.4 Status of action plans.......................................................................................................................................................60
Conclusions............................................................................................................................................................................................63
Recommendations ................................................................................................................................................................................64
Annexes...................................................................................................................................................................................................66
Annex I: Evaluation Statement of Work....................................................................................................................................66
Annex II: Data Collection Instruments.......................................................................................................................................74
Annex III: Works Cited..................................................................................................................................................................99
Annex IV: Women’s Survey Sampling Protocol.................................................................................................................... 101
Annex V: Completeness of Baseline Service Provider Data.............................................................................................. 102
Annex VI: Regression Tables ..................................................................................................................................................... 104
Annex VII: Initial Power Calcuations ....................................................................................................................................... 106
Annex VIII: Community Survey Data Collection Timeline ................................................................................................ 112
Annex IX: Comparison of Monthly Cases Recorded through Primary Data Collection and Monthly Cases 
Reported By The NPA ................................................................................................................................................................ 112
i 
ACKNOWLEDGEMENTS
An impact evaluation of this scope and rigor required the collaboration and contribution of many 
individuals and organizations. USAID’s Center of Excellence for Democracy, Human Rights and 
Governance played an essential role in promoting and ensuring a rigorous evaluation that not only tested 
impact, but tested two rival approaches to achieving impact. USAID/South Africa and the implementing 
partners of FPD, Soul City, and Sonke collaborated throughout the entire process and rolled out their 
programing in accordance with the randomization procedures required by the evaluation. With the 
support of the National Prosecuting Authority, TCC staff throughout the country collected data on 
utilization that was shared with the evaluation team and provides the backbone of this evaluation. Finally, 
a number of individuals participated in the design of the evaluation and the data collection. In short, this 
evaluation is the product of several years of hard work by many individuals. In particular, the authors 
would like to thank the following people and organizations for their generous support and contributions: 
Morgan Holmes and Vy Lam of USAID’s Center of Excellence for Democracy, Human Rights and 
Governance; Laura Berger and Paula Van Dyk at USAID/South Africa; Ria Schoeman and Gustaaf 
Wolvaardt at FPD; the staff of Soul City Institute and Sonke Gender Justice; the South Africa National 
Prosecuting Authority and TCC directors and staff; Mzikazi Nduna of the University of the Witwatersrand; 
Dr. Maximillian Kasy of Harvard University; Craig Carty; Aidan Connolly, Tronny Mawadzwa, Patrick 
Chiroro, Charles Katsande, William Muhwava, Charlene Flavell and all of the great field staff at Impact 
Research International; and Alison Miranda of Social Impact.
 
ii
ACRONYMS
APD Association of Persons with Disabilities
CAPI Computer Assisted Personal Interviewing
CLR Cluster Level Reliability
DiD Difference-in-Difference 
DOH Department of Health
DOJ/CD Department of Justice and Constitutional Development
DRG-LER Democracy, Rights, and Governance – Learning, Evaluation, and Research
DSD Department of Social Development
FAMSA Families South Africa
FPD Foundation for Professional Development
GBV Gender-based Violence
GoSA Government of South Africa
ICC Intra-class Correlation Coefficients
IE Impact Evaluation
IRI Impact Research International
ISSSASA Increasing Services for Survivors of Sexual Assault in South Africa
LGBTI Lesbian, Gay, Bisexual, Transgender, and Intersex
MDES Minimum Detectable Effect Sizes
MSCRT Multi-site Cluster Randomized Trial
NGO Non-governmental Organization
NPA National Prosecuting Authority
NVCS National Victims of Crime Survey
PCA Principal Component Analysis
PEP Post-exposure Prophylaxis
RCT Randomized Control Trial
SAECK Sexual Assault Evidence Collection Kits
SAPS South African Police Service
SGBV Sexual and Gender-based Violence
SI Social Impact
SOCA Sexual Offenses and Community Affairs
TCC
TVP
Thuthuzela Care Center
The Venus Project
USAID United States Agency for International Development
USAID/SA United States Agency for International Development/South Africa
VAO Victim Assistant Officer
VEP Victim Empowerment Programs
WHO World Health Organization
iii
TABLES
Table 1: Quantitative data sources by hypothesis and outcome indicator............................................................................ vii
Table 2: Quantitative data sources by hypothesis and outcome indicator.............................................................................. 6
Table 3: Assumptions for power calculations, precinct-level outcomes.................................................................................. 8
Table 4: Assumptions for power calculations, individual-level design ....................................................................................... 9
Table 5: Balance statistics for the reduced sample at baseline..................................................................................................27
Table 6: Regression results for impact of T1 on the logged avg. number of survivors using the TCC per month ....29
Table 7: Regression results for impact of T2 on the logged avg. number of survivors using the TCC per month ....31
Table 8: Number of women’s survey respondents by province...............................................................................................32
Table 9: Balance statistics between endline Treatment 1 and control samples ...................................................................33
Table 10: Respondent has heard of the TCC................................................................................................................................37
Table 11: Knowledge of TCCs by province...................................................................................................................................37
Table 12: Knowledge of TCC services............................................................................................................................................40
Table 13: Specific TCC knowledge...................................................................................................................................................41
Table 14: Number of SGBV survivors the respondent knows who have visited a TCC...................................................44
Table 15: Percentage of respondents indicating agreement with gender statements.........................................................47
Table 16: Number of professionals trained, surveyed at baseline and at endline ................................................................50
Table 17: Knowledge of TCCs..........................................................................................................................................................52
Table 18: Knowledge of TCC services............................................................................................................................................52
Table 19: Knowledge of more specific TCC services..................................................................................................................53
Table 20: Respondent recent practices with SGBV victims.......................................................................................................54
Table 21: Respondents’ contact and coordination with other service providers................................................................55
Table 22: Respondent attitudes toward SGBV..............................................................................................................................57
iv
FIGURES 
Figure 1: Treatment and control groups.......................................................................................................................................... vi
Figure 2: Results chain ........................................................................................................................................................................... 3
Figure 3: Treatment and control groups........................................................................................................................................... 5
Figure 4: TCC interview sites............................................................................................................................................................13
Figure 5: NGO interview sites ..........................................................................................................................................................14
Figure 6: Assessment of TCC ability to meet survivor need.....................................................................................................16
Figure 7: Independent assessment of TCC space.........................................................................................................................18
Figure 8: Independent assessment of TCC appearance ..............................................................................................................19
Figure 9: Independent assessment of TCC supplies ....................................................................................................................19
Figure 10: Average number of survivors presenting to each TCC per month.....................................................................26
Figure 11: Difference-in-difference estimate of the effect of community dialogues on TCC use....................................28
Figure 12: Difference-in-difference estimate of the effect of service provider training on TCC use .............................30
Figure 13: Endline respondent demographics................................................................................................................................34
Figure 14: Identification of service providers to victims of sexual assault .............................................................................35
Figure 15: Type of support services provided in the community.............................................................................................35
Figure 16: Respondents’ first response if victim of sexual assault............................................................................................36
Figure 17: Logistic regression results for effect of Treatment 1 on knowledge of TCCs.................................................39
Figure 18: Share of respondents who said they know about TCC services and location .................................................40
Figure 19: Respondent perception of community awareness of TCCs..................................................................................41
Figure 20: Share of respondents who heard ads/announcements about sexual assault in the last three months.......42
Figure 21: Respondent perception of crime problems at endline............................................................................................43
Figure 22: Share of respondents who know SGBV victims........................................................................................................44
Figure 23: Respondent perceptions of barriers to visiting the TCCs .....................................................................................45
Figure 24: Respondent perceptions of barriers to reporting to the police at endline........................................................46
Figure 25: Prevalence of victim-blaming attitudes among respondents at endline, by scenario.......................................48
Figure 26. Regression results for impact of community dialogues on attitudes toward SGBV........................................49
Figure 27: Respondent professions ..................................................................................................................................................51
Figure 28: Respondent age..................................................................................................................................................................51
Figure 29: Respondent perception of colleagues’ knowledge about the TCC .....................................................................53
Figure 30: Respondent comfort level talking with victims about SGBV .................................................................................55
Figure 31: Perceptions about SGBV reports and TCC usage....................................................................................................56
Figure 32: Perception about number of colleagues referring victims to the TCC ..............................................................56
Figure 33: Respondent level of victim-blaming ..............................................................................................................................58
Figure 34: Perception of whether victims would file a police report or go to the TCC...................................................58
Figure 35: Perception of police helpfulness....................................................................................................................................59
Figure 36: Main reasons for victims not filing a police report or going to the TCC...........................................................59
Figure 37: Perception of TCC helpfulnes)......................................................................................................................................60
Figure 38: Status of SGBV project selection ..................................................................................................................................61
Figure 39: Status of activities within SGBV projects....................................................................................................................61
Figure 40: Reasons for not planning an SGBV project at their work facility .........................................................................62
v 
EXECUTIVE SUMMARY
INTRODUCTION
This report presents findings from the baseline data collection for USAID/South Africa’s (USAID/SA) 
Increasing Services for Survivors of Sexual Assault in South Africa (ISSSASA) Program. The objective of 
this program is to improve service provision and community awareness of services for survivors of sexual 
assault in South Africa, which struggles with one of the highest rates of gender-based violence in the world 
(Gender Links and Medical Research Council, 2011). The Government of South Africa’s (GoSA) fight 
against sexual and gender-based violence (SGBV) is spearheaded by the Sexual Offenses and Community 
Affairs (SOCA) unit of the National Prosecuting Authority (NPA) within South Africa’s Department of 
Justice and Constitutional Development (DOJ/CD). USAID has worked with the NPA/SOCA since 1999 
to establish the Thuthuzela Care Center (TCC) model.1 TCCs provide a comprehensive portfolio of 
services to survivors of SGBV, including emergency medical care, psychosocial counseling, post-exposure 
prophylaxis (PEP), HIV testing and counseling, and assistance with case reporting and court preparation, 
in an integrated and victim-friendly manner. The TCC model seeks to streamline the care process for 
SGBV survivors by establishing effective linkages between various service providers and government 
stakeholders, and to improve legal services by reducing time-to-court and increasing the conviction rate. 
This impact evaluation (IE) is a rigorous study of the effectiveness of two distinct intervention approaches 
to increase the rates of SGBV survivor reporting, follow-through with services, and public awareness and 
understanding of SGBV and resources available to survivors. The first of these interventions is a demand￾side intervention implemented by Soul City Institute (Soul City) and Sonke Gender Justice (Sonke). While 
the implementers had developed a suite of demand-side outreach activities, this evaluation focused 
specifically on testing the effect of community dialogues designed to provide information about SGBV and 
TCCs. The second is a supply-side intervention implemented by the Foundation for Professional 
Development (FPD) which included multi-disciplinary trainings for service providers in the TCC referral 
and care networks. 
This IE aims to provide evidence about the effectiveness of supply-side versus demand-side outreach 
activities for improving service awareness in South Africa, and in similar contexts. The outcomes of this 
IE are expected to be highly informative for both the academic and development communities, and for 
stakeholders working to address SGBV in South Africa. 
EVALUATION QUESTIONS AND DESIGN
This IE utilizes a randomized control trial (RCT) design to assess the independent effects of each of the 
two interventions. This experimental evaluation design enables estimation of the average effect of each of 
the two interventions on the outcomes of interest by comparing communities that received either of the 
interventions to those that did not. While the implementers of this intervention requested that the IE be 
conducted with the treatment defined as a holistic suite of program activities, as this was the intervention 
designed at the procurement stage, it was ultimately decided that the IE would study the impact of the 
components separately to better understand what approach works best and why. Unfortunately, due to 
sample size and resource constraints, it was not possible to test the combined effect of both interventions 
administered simultaneously and test for independent program effects. In addition to separating out 
program components, there were several ways in which the intervention had to be accommodated to 
permit an impact evaluation. 
 
1 Thuthuzela is a Xhosa word meaning “to comfort.” 
vi
The IE was designed to address the following evaluation questions: Do the interventions:
1. Increase utilization of TCCs?
2. Increase public knowledge about SGBV and TCCs?
3. Reduce stigma associated with SGBV?
At the start of this IE, there were 51 TCCs operating in South Africa.2 Three police precincts around each 
of the 51 TCCs were selected to participate in the IE, with each of the three precincts randomly assigned
to one of the three study groups—control, demand-side Treatment 1, or supply-side Treatment 2.3 The 
communities selected for the evaluation were thus distributed as follows: 50 communities received the 
demand-side community dialogues outreach program, 50 received the supply-side service provider 
training, and the remaining 50 received no programming and serve as the control group.
Figure 1: Treatment and control groups
The measurement of evaluation outcomes for this IE relies on administrative and government secondary 
data, and quantitative and qualitative data collected as part of the evaluation. At the precinct level, the 
evaluation team worked in collaboration with the NPA and the TCCs to collect precinct specific 
administrative data on the use of TCC services. As suggested by Table 1, by comparing these TCC records 
across the treatment and control precincts, the evaluation tests whether, and by how much, each of the
two interventions increased use of TCC services (Hypothesis 1). These quantitative data were
complemented by qualitative data collected through interviews. A total of 94 baseline interviews were 
conducted across all 51 TCC catchment areas. Interviewees included senior staff from 51 TCCs,4
representatives of 40 local non-governmental organizations (NGOs)—many of which were characterized 
as Victim Empowerment Programs (VEPs)—and three independent counsellors (non-NPA employed) who 
assist with survivor case management. Follow-up interviews were also conducted with each TCC as part 
of endline data collection. Forty-eight endline interviews were conducted with TCC senior staff, while 
three TCCs were unwilling to participate in the endline.
 
2 Since the beginning of this IE, three more TCCs were constructed. These are not included in this evaluation. 
3 Because two TCCs reported working with fewer than three police precincts (one TCC works with one precinct and another 
works with two), the number of police precincts per treatment arm was reduced to 50. 
4 One TCC interview was excluded from analysis due to discrepancies between the researcher’s original notes and the final 
transcription notes. See section 2.3.5 of this report for more detailed discussion.
vii
Table 1: Quantitative data sources by hypothesis and outcome indicator
Hypothesis Outcome Indicator Data Source
Precinct-level 
H1: Interventions cause an 
increase in reporting of SGBV 
and utilization of TCC services
SGBV reporting Police & TCC records, 
Supplemental Intake Form
Utilization of TCC services TCC records, Supplemental 
Intake Form
Follow-through with TCC services TCC records, Supplemental 
Intake Form
Individual-level
H2: Interventions cause an 
increase in knowledge of SGBV 
and TCCs
Community knowledge of SGBV Women's survey
Service providers' knowledge of SGBV Service provider survey
Community knowledge of TCCs Women's survey
Service providers ' knowledge of TCCs Service provider survey
H3: Interventions cause an 
improvement in attitudes 
toward SGBV
Community perceptions of SGBV Women's survey
Service providers' perceptions of SGBV Service provider survey
To test whether the interventions increased knowledge of SGBV and TCCs (Hypothesis 2) and/or changed 
attitudes toward SGBV (Hypothesis 3), the evaluation collected baseline and endline data through surveys 
in communities where dialogues were held (Treatment 1) and comparison communities, and among 
service providers participating in trainings (Treatment 2). Specifically, the evaluation tested whether 
community outreach campaigns increased community knowledge of and attitudes toward SGBV through 
a survey of 1,500 women in Treatment 1 and control areas at baseline and endline in all nine provinces in 
South Africa where TCCs operate. The evaluation also tested whether service providers’ knowledge, 
attitudes, and self-reported behavior changed following the trainings through a pre- and post-training 
survey of participating service providers. Of the total 1,908 training participants, 1,789 (94 percent) 
participated in the evaluation at baseline and 1,347 (71 percent) participated at endline.5 While the 
response rate among training participants was high, not all participants completed the survey in its entirety, 
so there are some missing data on most survey questions. 
A number of limitations to this IE should be acknowledged. First, TCC compliance in completing 
supplemental intake forms was lower than anticipated. Three TCCs did not submit any supplemental 
intake forms to the evaluation team. Of the remaining 47, 31 reported using intake forms with 90-100 
percent of survivors. Moreover, aggregate numbers reported by the NPA are notably higher than the 
number of cases recorded in the supplemental intake forms (see Annex IX). However, this limitation does 
not appear to bias the IE findings as underreporting would affect both control and treatment areas. Second, 
the theory of change linking the community dialogues, which only targeted a limited number of participants, 
to increases in knowledge and changes in attitudes in the community as a whole was not strong enough; 
the results could have been different had a more robust demand-side intervention been tested. Third, 
respondents were sampled based on treatment and control locations and survey data were not weighted, 
thus household survey data cannot be considered nationally representative of women in South Africa, nor 
representative at the provincial level. Fourth, the data collection team experienced some barriers to entry 
in four police precincts near Cape Town due to high levels of violence and crime coupled with racial 
tensions in those areas. To conduct data collection in this area, the evaluation team partnered with a 
highly localized data collection firm with experience working in these areas to survey households in those 
precincts, and paper-based data collection was used in these areas due to the security challenges. Finally, 
while most TCC staff were receptive to qualitative interviews, several refused to participate at some point 
during the study. It is likely that some respondents did not feel they could speak freely, despite assurances 
 
5 Excludes participants from Margate precinct in KwaZulu-Natal province. This precinct was dropped from the evaluation due to 
delays in implementation. 
viii
that the audio recording and interview notes would be kept confidential. 
FINDINGS AND CONCLUSIONS
TCC capacity concerns 
Qualitative data on TCC capacity collected via in-depth, targeted interviews yielded important findings 
about the challenges TCCs face.
• Variation in accessibility: Twenty-three TCCs were open Monday through Friday while 20 
were open seven days a week. Twenty-one TCCs reported operating 24 hours a day with support 
from non-TCC staff, typically NGO employees. Operating hours for TCCs that were not open 
around the clock ranged from 7:00 AM to 7:00 PM, with the majority open between standard 
business hours (8:30 AM to 4:30 PM). Most TCCs that did not operate on a 24-hour schedule 
referred survivors to external partners for services after regular business hours. Some after-hours 
services were available at 34 TCCs, 33 of which received after-hours support from external 
partners.
• Under-resourced: Even with NGO employees supplementing TCC staff, many TCCs reported 
being chronically understaffed. In thirty TCCs, at least one staff position was not filled at the time 
of the interview, and ten other TCCs had one or more unfilled staff positions in the months 
preceding the interview. 
• Concerns over capacity: Few TCC Site Coordinators expressed high confidence in their TCC’s 
capacity and ability to meet survivors’ needs successfully. Respondents were asked to rate their 
TCC’s overall capacity on a four-level scale ranging from “completely adequate” to “not at all 
adequate.” Only one respondent rated their TCC as fully adequate. TCC respondents described 
their TCCs as having limited resources, supplies, staffing, and facilities as barriers to serving 
survivors. 
• Barriers to utilization: Interview respondents cited numerous factors that pose barriers to 
TCC utilization. Respondents observed that survivors were not aware of TCC services and had 
misperceptions about the type of services provided. Many TCCs were quite difficult to locate, 
even for experienced research staff. Despite having coordinates for each TCC, the research team 
was unable to find 31 of 50 sites without asking for directions. While this appears to have been 
done by design to prevent perpetrators from finding the TCCs, this is likely to severely undermine 
TCC utilization. Insufficient access to transportation was also cited as a barrier.
• Important collaboration with NGOs: TCCs typically partner with NGOs that support their 
work with survivors, as well as extend and integrate services throughout the community. Nearly 
all NGO representatives reported favorable, productive working relationships with TCCs and 
understanding of TCC resource constraints.
No impact of community dialogues and service provider training on TCC utilization
Supplemental intake data collected from the TCCs show that TCC use is low, with a median of 12.7
survivor cases per month per TCC across the evaluation period.6 Both the full and reduced samples of 
control and treatment precincts were balanced at baseline on the dependent variable and the relevant 
control variables. Difference-in-difference regression analyses show no significant impact of the community
dialogues or service provider training on TCC utilization. Statistically significant explanatory variables for 
 
6 The median of 12.7 cases per month is on the restricted sample of 24 TCCs that were estimated to have collected data 
consistently with at least 90% of survivors. As a likely result of inconsistent data collection, the average of the full sample is 
somewhat lower. The NPA reports a notably higher median of 22.5 cases per month among these 24 TCCs (see Annex IX). 
ix
TCC use include population size of the precinct, rate of sexual offences, and rate of murders. 
No impact of community dialogues on community knowledge and attitudes
Logistic regression was used to test the impact of community dialogues on awareness of TCCs. The result
reveal that knowledge of TCCs in sampled communities was statistically similar at baseline and endline. 
Moreover, knowledge levels between treatment and control communities are statistically equivalent,
indicating that there was no positive treatment effect of the community dialogues on knowledge of TCCs 
or services available to survivors of SGBV. This finding suggests that community dialogues by themselves 
are insufficient to have a multiplicative effect and influence the larger community in which they are held. 
This is perhaps not surprising and suggests the need to test the effect of a more robust demand side 
intervention, consistent with the full ISSSASA program, but focused locally and more intensely targeted
to reach a critical mass of community members with repeated messaging. 
This is unfortunate, as the survey demonstrates that knowledge of the TCCs is low (around 18 percent) 
and perceived to be the greatest barrier to TCC use. Instead, of treatment, province is the strongest and 
most significant predictor of familiarity with TCCs, with knowledge lowest among women in Gauteng and 
highest in Eastern Cape. One potential explanation for this variation is that Thuthuzela is a Xhosa word, 
which would be most familiar in Xhosa speaking areas such as the Eastern Cape. Sexual assault prevalence 
and perception of sexual assault as a problem are also significant in predicting knowledge of TCCs, and
women who reported knowing someone who has been raped or sexually assaulted were more likely to 
know of the TCC. Respondents with low tolerance for violence toward women were also slightly more 
likely to know of TCCs. Women who are older, more educated, and with higher household income are 
more likely to have heard of TCCs. 
Regression analysis was also used to test whether women in treatment precincts scored higher on an 
attitudinal SGBV index than women in control precincts after the intervention. The survey shows most 
respondents have normatively desirable attitudes toward gender roles; however, some respondents still 
engage in victim blaming and there is still evidence to suggest that social stigmas represent a barrier to 
TCC usage. The regression estimate shows a slight improvement in attitudinal scores from baseline to 
endline, in the treatment group relative to the control group; however, this relationship is not statistically 
significant. Variables that are statistically significant and positively associated with more normatively 
desirable SGBV attitudes scores are: age, Zulu, higher income, and knowing a woman or girl who has been 
raped or sexually assaulted. 
Positive impact of training on service providers’ knowledge and attitudes
Service providers attending the FPD training completed a pre- and post-training survey to measure 
changes in knowledge, attitudes, and self-reported practices. At endline, service providers were 
statistically significantly more likely to report knowing what services were available for survivors of SGBV 
in their communities, what services were provided by the TCCs, and the location of TCCs. 
The service provider survey also included a series of questions designed to assess attitudes toward gender 
equity norms. At baseline, many professionals revealed moderate levels of victim-blaming. A third of 
respondents believed that women provoke rape by their appearance or behavior and nearly half indicated 
that the extent of a woman’s resistance should be the major factor in determining if a rape has occurred. 
Statistically significant positive changes were observed on seven of nine measures of victim-blaming and 
x 
tolerance of violence against women, one measure showed a negative change,7 and one measure remained 
at similar levels observed at baseline.8 
The evaluation also sought to measure whether Treatment 2 training participants adopted behavior 
changes in how they interact with SGBV survivors by asking what actions they had taken to assist SGBV 
survivors in the last 60 days. At endline, no positive changes were observed over time in most of the 
practices included in the survey, although more respondents reported taking someone to a TCC and 
informing a victim about TCC resources, with the latter measure being statistically significant. 
RECOMMENDATIONS
Based on the findings and conclusions from this impact evaluation of the ISSSASA activity, the evaluation 
team recommends that USAID, the NPA, and other stakeholders consider the following areas of 
intervention to continue developing the system of care for survivors of sexual assault. 
• Support TCC capacity. TCCs reported various areas of need for support across staffing, 
facilities, and supplies. In a resource-constrained environment and given the relatively low levels 
of TCCs use justification for added staffing or other resources may be a challenge. On-call staff 
that are available 24 hours a day seven days a week, for example, as opposed to staff sitting at the 
TCC around the clock may be a more reasonable use of resources, especially in areas with lower
TCC use. Continuing to encourage partnerships with NGOs and community organizations, which 
we find to be providing important complementary services and filling resource voids, is another 
promising area for intervention. The findings above indicate some changes that would be effective 
while expending minimal resources. For example, the policy of keeping TCCs’ locations somewhat 
secret undermines utilization and should be reconsidered. • Test the effects of a more broad-based, social marketing approach to raising 
awareness about TCCs and SGBV. Given continued low awareness of the TCCs within 
surrounding communities and continued apprehension about reporting SGBV, more intensive 
outreach efforts should be tested. The broader ISSSASA program includes a television drama 
series, a school based intervention engaging children in grade 7, a radio public service 
announcement campaign, and a digital and social media campaign. Efforts to increase awareness 
and change attitudes should build on best practices in social marketing and be designed to reach 
and influence a wider audience with sustained and repeated messaging via a “saturation” approach. 
Such an approach would be amenable to a follow-on impact evaluation using the same survey 
methodology applied here and could even be focused on communities with the greatest perceived 
need. The intervention would require a stronger theory of change that would be likely to produce 
a change detectable in a random sample of women in targeted communities. As one reviewer 
noted in comments, “…social change at the community level often occurs gradually and over a 
more protracted period of time and with repeat exposure as opposed to a one-off exposure to 
an intervention.”
Furthermore, given that the TCCs report limited to no direct community outreach activities at 
present, the evaluation team recommends that TCC staff engage directly with communities, 
thereby building connections with schools, clinics, police, community organizations, and 
community members themselves. These direct relationships have the potential to improve the 
status of the system of care, build trust, and increase awareness. Data from this evaluation shows 
that few TCC staff were available to participate in community dialogues or service providers’ 
 
7 This negative change could be explained by the wording of the question, which could have confused respondents. 
8 Negative change was observed on “women do not provoke rape by their appearance or behavior,” and no change was measured 
for “women often claim rape to protect their reputations.”
xi
trainings due to their responsibilities at the TCC or lack of transportation. Human resource 
limitations remain a challenge at the TCCs; however, centralized staff could help coordinate 
outreach efforts. Where it is not possible to send TCC staff into communities for outreach, 
consider technology, social media, or radio as a means for TCC staff to engage with communities. 
• Adopt a train-the-trainer approach to community dialogues. Community dialogues are 
an attractive option towards meeting the goals to decrease SGBV incidence and increase support 
for survivors, as they allow for a more in-depth discussion of SGBV issues, which is likely necessary 
to change attitudes and reduce stigma. Nonetheless, these dialogues are only able to reach a small 
number of participants, and we did not find evidence that they were effective in achieving the 
objectives tested in this evaluation. As such, they should either be used strategically as part of a 
“saturation” strategy in targeted communities, or they should be scaled-up dramatically. As part 
of the latter approach, we would recommend adopting a train-the-trainer or promoter-based 
approach. There is some precedent for this, as Soul City has been working to establish clubs of 
young women who could lead their own dialogues. Participants should be provided with the skills 
and materials to encourage information proliferation while ensuring fidelity to the established
approach. 
• Formalize SGBV training for police and other professionals in the system of care. 
Participants in the Integrated Management training were recruited via open invitations to service 
providers at targeted institutions. As such, participants self-selected into the program and many 
had a strong interest in SGBV or were already working on SGBV. In fact, at baseline, 80 percent
of service providers already knew of SGBV services and 65 percent had informed a victim about 
the TCCs in the last 60 days. This could explain why the intervention did not increase referrals 
from this group. It seems probable that a training could be most beneficial if it were provided to 
service providers who likely have contact with SGBV survivors but are not knowledgeable about 
such issues nor the services that exist to address them. These individuals would be less motivated 
to attend a three-day training, but a shortened training could be offered to a wider audience of 
service providers with a heavy emphasis on follow-up action. 
Police would be an obvious target for such training; however, USAID/South Africa reports that it 
is difficult to work with the South Africa Police Service in such capacity due to Leahy Law 
requirements for training. Training nurses in making referrals would be an attractive alternate
target group, as these professionals often interact with survivors. While nurses have many 
demands on their time, they might be amenable to such a training because the TCCs offer a 
support service that can make their jobs easier. If this were to occur, we would recommend 
conducting the same pre- and post-survey. Some of the knowledge and attitude gains might be 
lower with a less motivated group; however, a well delivered training could lead to greater 
awareness throughout the service provider community and increase referral sources.
• Use these data to inform future programming: The survey data demonstrate a number of 
knowledge gaps, important perspectives, and potentially problematic attitudes that could help 
USAID, the NPA, and the IPs improve their programming. For example, while the survey reveals 
generally progressive attitudes, it also shows that many individuals think there are limits on a 
woman’s right to refuse sex and there is evidence of victim blaming in certain situations. These 
statistics should be incorporated into curriculum, discussed, and addressed. 
1 
INTRODUCTION
This report presents the final results of the impact evaluation of USAID/South Africa’s (USAID/SA) 
Increasing Services for Survivors of Sexual Assault in South Africa (ISSSASA) Program, under the 
Democracy, Rights, and Governance – Learning, Evaluation, and Research (DRG-LER) activity. The 
ISSSASA project is a five-year, $10 million cooperative agreement managed by USAID/SA’s Democracy, 
Human Rights, and Governance Office that began in June 2012 and is scheduled to run through June 2017. 
The objective of this program is to improve service provision and community awareness of services for 
survivors of sexual assault by expanding TCC services, as well as raising public awareness of TCCs, TCC 
services for survivors of sexual assault, and sexual and gender-based violence (SGBV) in general. 
1.1 EVALUATION PURPOSE
Jointly commissioned by USAID/SA and the Learning Team at USAID’s Center for Excellence in 
Democracy, Human Rights and Governance, this IE rigorously evaluates the effectiveness of two distinct 
intervention approaches to improving SGBV survivor reporting and increasing public awareness and 
understanding of SGBV and resources available to survivors. Results from this study are intended to help 
inform USAID and implementing partners on effective approaches to addressing challenges in TCC and 
SGBV awareness, reporting, and service follow-through, and will also provide general information on the 
function and role of the TCCs in providing services to survivors of sexual assault in South Africa. 
This IE aims to provide evidence about the effectiveness of two different approaches—supply-side versus 
demand-side outreach activities—to improve TCC service awareness in South Africa, and in similar 
contexts. The outcomes of this IE are expected to be highly informative for both academic and 
development communities, and for stakeholders working to mitigate SGBV in South Africa. The findings 
of this IE also have implications for funding decisions bearing on accountability and other development 
objectives.
1.2 PROJECT BACKGROUND
Sexual and gender-based violence is a pervasive global health and development problem with substantial 
physical, social, and economic consequences. SGBV is committed by both intimate partners and strangers, 
and while both men and women experience and perpetrate SGBV, it is most commonly perpetrated by 
men against women. In all its forms, SGBV is a human rights violation rooted in gender inequality, 
patriarchal social norms, and rigid gender roles that equate masculinity with violence. SGBV is linked to 
numerous health problems, including physical injuries, psychological trauma, unwanted pregnancy, adverse 
pregnancy outcomes, sexually transmitted infections (including HIV), issues with contraception and 
abortion, and increased mortality.
South Africa has one of the highest rates of SGBV in the world. In 2014/2015, the sexual offences rate in 
South Africa was 99.3 per 100,000 people nationwide (Institute for Security Studies, 2015). The Statistics 
South Africa’s National Victims of Crime Survey (NVCS) found that most victims of any type of assault 
knew their perpetrator, who were either from the same community (34.2 percent), a spouse or lover 
(16.8 percent), or a relative (9.2 percent) (Institute for Security Studies, 2015). Moreover, the NVCS also 
shows that the rate of reporting assaults to the police has been declining in South Africa in recent years. 
Best estimates indicate that at most only one in nine rapes are reported to authorities (Jewkes and 
Abrahams, 2002). Together, these signal that the true rate of sexual assault in South Africa is likely much 
higher than official crime statistics indicate. 
One in four adult South African women report having experienced sexual and/or physical intimate partner 
violence in their lifetimes (Shai and Sikweyiya, 2015). Over half of all female homicides in South Africa are 
2 
committed by an intimate partner—six times higher than the global average—and women of color and 
women ages 14 to 44 are especially vulnerable to mortality from intimate partner violence (Abrahams et 
al., 2009).
A 2011 study found that 42 percent of men disclosed having perpetrated intimate partner violence and 
almost 28 percent of South African men had raped a woman, whether an intimate partner, acquaintance, 
or stranger (Jewkes et al., 2011). Over half of those men committed rape on multiple occasions, and 75
percent perpetrated their first rape before age 20. In addition to perpetrating sexual violence, young 
people are particularly likely to be victimized; 60 percent of survivors presenting at TCCs are under 18, 
and 40 percent of the survivors are under the age of 12 (South African government, 2013). 
Although the scope of SGBV in South Africa has been increasingly documented, the subject remains under￾studied because many survivors do not report or discuss their experiences. Stigma, shame, and fear, as 
well as financial and emotional dependency on perpetrators, often deter survivors from reporting SGBV 
or seeking help. Further, current social structures tend to embed permissive patriarchal norms, condone 
sexual assault, and even stigmatize survivors and those who utilize SGBV services.
The GoSA fight against SGBV is spearheaded by the Sexual Offenses and Community Affairs (SOCA) unit 
of the National Prosecuting Authority (NPA) within South Africa’s Department of Justice and 
Constitutional Development (DOJ/CD). USAID has worked with the NPA/SOCA since 1999 to establish 
the Thuthuzela Care Center (TCC) model. TCCs provide a comprehensive portfolio of services to 
survivors of sexual violence, including emergency medical care, psychosocial counseling, post-exposure 
prophylaxis (PEP), HIV testing and counseling, and assistance with case reporting and court preparation in 
an integrated and victim-friendly manner. The TCC model seeks to streamline the process for SGBV 
survivors by establishing effective linkages between various service providers and government 
stakeholders, and to improve legal services by reducing time-to-court and increasing the conviction rate.
Improperly or inadequately trained responders and service providers can further victimize SGBV 
survivors, who, in particular, require sensitivity and attention to privacy, confidentiality, and security. TCCs 
work to avoid possible secondary victimization, which can take the form of survivors being blamed or 
disbelieved, having to give their statements multiple times, or being forced to exhibit injuries or recount 
experiences in open areas of police stations. Even for SGBV survivors who report their experiences and 
receive services and medical care, successful prosecution of perpetrators is rare. In 2012/2013, only seven
percent of reported sexual offense cases resulted in conviction (Gibbs et al., 2014). The NPA is currently 
supporting research to better understand the challenges to successful prosecution of perpetrators and 
the best ways to overcome them. One key challenge is the collection, analysis, and presentation of medico￾legal evidence, which includes genital and non-genital injuries and DNA evidence. Evidence of injuries are 
currently recorded on J88 forms at TCCs, which are completed by doctors or forensic nurses, and DNA 
evidence is collected through Sexual Assault Evidence Collection Kits (SAECK). Physical evidence has been 
shown to be a strong factor in successful sexual violence prosecutions (Gibbs et al., 2014), but its use has 
been inhibited by improper collection and handling of evidence and by victims not always coming forward 
right away.
While improving the quality of care is essential, it is imperative that victims actually access and utilize 
SGBV services. SGBV continues to carry significant stigma and many potential victims remain unaware of 
TCCs and their services. A formative research effort by the NGO Soul City identified significant barriers 
to the access and use of TCC services (Soul City Institute, 2013). The study conducted interviews with 
fifteen focus groups of eight to 12 participants each, spanning rural, semi-urban, and urban areas of five 
provinces to assess the general public’s knowledge of sexual assault, TCCs and TCC services. Three of 
the main barriers identified by the study included: (1) shame and stigma associated with sexual assault; (2) 
lack of knowledge about the TCCs and TCC services; and (3) poor institutional support for the TCC 
referral system, i.e., the police and school teachers. This impact evaluation (IE) focuses on evaluating the 
effectiveness of interventions to alleviate the second and the third barriers to TCC utilization. This focus 
3 
area was selected in consultation with the stakeholders and the USAID/SA Mission staff, taking into 
account the IE design and implementation constraints.
This IE evaluates two specific approaches to increasing public awareness about and increased use of TCCs, 
representing only a portion of the USAID-funded ISSSASA program: 
Demand-side intervention, multi-media community dialogues: This demand-side intervention provides 
information (e.g., flyers, posters) about TCCs and TCC services to the local communities, through a 
community dialogue format, hosted by two CSOs—Sonke Gender Justice and Soul City. The dialogues 
seek to educate community members about sexual assault and other SGBV issues, and dispel common 
misperceptions of TCC services. Soul City hosts dialogues for women and girls, while Sonke Gender 
Justice hosts dialogues for men and boys.
Supply-side intervention, multi-disciplinary training programs: This supply-side intervention provides training 
for the professional service providers in TCC referral and care networks. Within each community, 
approximately thirty multi-disciplinary professionals, including police officers, teachers, social workers, 
health professionals, NGO representatives, and TCC staff attend a training conducted by FPD. Participants 
are trained on the legal framework and support standards for provision of services to survivors of SGBV, 
child protection, and court/litigation preparation. Roving teams provide follow-up with the trainees at 
their home institutions. The results chain for each of the two interventions is presented in Figure 2 below. 
The theory of change posits that the demand-side community dialogues and supply-side service providers 
training will have a direct effect on increasing awareness of TCCs and the associated services, therefore 
potentially increasing reporting of SGBV and utilization of TCC services, and improving the rate of follow￾through with support services and criminal cases.
 
Activities Outputs Outcomes Goals
Demand-side 
Outreach
• Community 
dialogues
• Community 
member 
attendance & 
participation
• Flyers & posters 
disseminated 
• Increased 
knowledge about 
SGBV, TCCs, 
and TCC 
services
• Reduced stigma 
associated with 
SGBV and 
accessing TCCs
• Increased TCC 
utilization
• Reduced risk 
factors and 
incidence of 
SGBV
• Increased follow￾through with 
criminal cases
Supply-side
Outreach
• Multi-disciplinary 
service 
providers’ 
trainings
• Trainee 
attendance & 
participation
• Follow-up work 
conducted
Figure 2: Results chain
4 
EVALUATION DESIGN 
This IE utilizes an RCT design to assess the effectiveness of each of these two interventions. This 
experimental evaluation design enables estimation of the independent effect of each of the two 
interventions on the outcomes of interest by allowing effective comparison of communities that received 
either intervention to those that did not. As such, the RCT design permits causal identification by reducing 
selection bias and other endogeneity problems and by controlling for confounding variables (Angrist and 
Pischke 2010; Banerjee and Duflo 2008; Duflo, Glennerster, and Kremer 2006), permitting inference about 
the effectiveness of the supply-side versus the demand-side approach. 
It should be noted that the implementers of this intervention expressed concern about testing the 
independent effects of these two interventions, and had preferred that the IE focus on the suite of program 
activities intended to work together. Interventions like the ISSSASA program are often holistic programs 
with a number of complementary elements. While it is possible to conduct an evaluation of a multipronged 
approach, it becomes difficult to interpret the implications of the findings. For example, if the impact is 
found to be positive and significant, should the whole approach be scaled up or exported or are the 
advances driven by a particular element? Alternatively, if no impact is found, are there parts of the 
intervention that might be working? To avoid this challenge, USAID’s Center of Excellence on Democracy, 
Human Rights, and Governance has favored evaluating specific aspects of an intervention and attempting 
to measure independent effects, and this is the approach that was adopted in this evaluation. The risk with 
such an approach, however, is that individual activities might be insufficient to produce an effect on their 
own even if they might be an effective complement to other efforts. In this case, the theory of change 
linking the community dialogues, which only targeted a limited number of participants, to increases in 
knowledge and changes in attitudes in the community as a whole was not strong enough; the results could 
have been difference if a more robust demand-side intervention were tested. Furthermore, in an ideal 
scenario, the evaluation would have included a third treatment arm to measure the effectiveness of 
combining robust demand-side and supply-side approaches; unfortunately, due to sample size and resource 
constraints, this was not possible. In addition to separating out program components, there were several 
ways in which the intervention had to be accommodated to permit an impact evaluation. 
2.1 EVALUATION QUESTIONS AND HYPOTHESES
This impact evaluation was designed to address the following evaluation questions:
Do the interventions:
1. Increase utilization of TCCs?
2. Increase public knowledge about GBV and TCCs?
3. Reduce stigma associated with SGBV?
The resulting hypotheses9 are: 
H1: Interventions will have a positive effect on the reporting of SGBV and on take-up of TCC 
services.
H2: Interventions will have a positive effect on the community and professionals’ knowledge of 
 
9 Discussions with the implementing partners and USAID revealed concerns over the ability of the intervention to measurably 
impact stigma associated with SGBV or follow-though with criminal cases. Accordingly, the evaluation team considers Hypotheses 
1 and 2 as primary, and Hypothesis 3 as secondary. 
5 
SGBV, TCC presence and TCC services.
H3: Interventions will improve community and professionals’ attitude toward SGBV. 
To test these hypotheses for each of the interventions, 150 police precincts around 51 TCCs were 
randomly selected into one of three groups: (1) a demand-side treatment group, consisting of communities 
in which multi-media community dialogues were conducted, (2) a supply-side treatment group, from which 
multi-disciplinary service providers were recruited to participate in a training program, and (3) a control 
group that did not receive any SGBV related intervention as part of the ISSSASA Program.10 A schematic 
diagram of this design is shown in Figure 3. 
Figure 3: Treatment and control groups
This IE utilizes two layers of measurement: the police precinct level and the individual level. At the precinct 
level, the evaluation team has worked in collaboration with the NPA and the TCCs to collect precinct 
specific administrative data on use of TCC services. By comparing these TCC records across the 
treatment and control precincts (shown in Table 2), the evaluation team is able to test which if any of the 
two interventions increases use of TCC services (Hypothesis 1). These quantitative data are 
complemented by qualitative data collected through interviews with TCC staff and supporting NGOs. 
 
10 Because two TCCs reported working with fewer than three police precincts (one TCC works with one precinct and another 
works with two), the number of police precincts per treatment arm was reduced to 50. 
6 
Table 2: Quantitative data sources by hypothesis and outcome indicator
Hypothesis Outcome Indicator Data Source Precinct- level
H1: Increase in reporting of GBV 
and utilization of TCC services
SGBV reporting Police & TCC records, 
Supplemental Intake Form
Utilization of TCC services TCC records, 
Supplemental Intake Form
Follow-through with TCC services TCC records, 
Supplemental Intake Form
Individual-level
H2: Increase in knowledge of 
GBV and TCCs
Community knowledge of SGBV Women's survey
Service providers' knowledge of SGBV Service provider survey
Community knowledge of TCCs Women's survey
Service providers ' knowledge of 
TCCs
Service provider survey
H3: Improvement in attitudes 
toward SGBV
Community perceptions of SGBV Women’s survey
Service providers’ perceptions of 
SGBV
Service provider survey
To answer Hypotheses 2 and 3, the evaluation team collected baseline and endline surveys in communities 
where dialogues were held (Treatment 1) and comparison communities, and among service providers 
participating in trainings (Treatment 1). Specifically, we tested if community outreach campaigns increase 
community knowledge of and attitudes toward SGBV through a survey of 1,500 women in Treatment 1
and control areas at baseline and endline in all nine provinces in South Africa where TCCs operate.11 The 
evaluation team also tested whether service providers’ knowledge, attitudes, and self-reported behavior 
changed as a result of trainings through a pre- and post-training survey with participating service providers. 
Of the total 1,908 training participants, 1,789 (94 percent) participated in the evaluation at baseline and 
1,347 (71 percent) participated at endline.12 
2.2 SAMPLING, RANDOMIZATION, AND POWER CALCULATIONS
This evaluation was designed as a multi-site cluster randomized trial (MSCRT). The sampling approach is 
bound by the total number of sites (TCCs) and the total number of geographic clusters (police precincts 
served by each TCC). 
2.2.1 Precinct level 
At the start of this IE, there were 51 TCCs operating in South Africa;13 with each TCC serving from one
to 18 police precincts and the majority of TCCs serving three to six police precincts. The treatment and 
control groups were balanced to minimize bias during the sampling and randomization process. Three 
police precincts around each of the 51 TCCs were selected to participate in the IE.14 A sophisticated 
balanced assignment technique developed by Maximillian Kasy was applied to select three precincts within 
 
11 This IE was initially designed to evaluate knowledge and attitude changes among men and women. However, due to resource 
constraints, the final design opted to focus on the female community members only, to maximize the sample size of female 
respondents and maximize the MDES for that group. We recognize that men’s knowledge and attitudes are an important area 
for future studies since prevailing norms among men are critical in understanding and decreasing SGBV. 12 Excludes participants from Margate precinct in KwaZulu-Natal province. This precinct was not included in the evaluation due 
to delays in implementation. 
13 Since the start of this IE, three more TCCs have been constructed, which are not included in the evaluation. 14 Two TCCs reported working with fewer than three police precincts (one TCC worked with two precincts and another TCC 
with one precinct), reducing the number of police precincts per treatment arm to 50. 
7 
each TCC catchment area and to assign each precinct to one of three groups: control (no treatment), 
demand-side Treatment 1, or supply-side Treatment 2.15 
The balancing technique to select precincts optimally minimizes the expected squared error of estimators 
of treatment effects, based on a set of precinct covariates that balance on several characteristics. These 
included variables from census data and police crime statistics for each of the precincts likely correlated 
with sexual assault prevalence, specifically: population size, sexual assault rate, homicide rate, and 
percentage of the population that is rural.16 The regression specification was: 
𝑌𝑌𝑖𝑖𝑖𝑖 = 𝛿𝛿𝑡𝑡 + 𝛿𝛿 + 𝛸𝛸𝑖𝑖𝑖𝑖 ⋅ 𝛽𝛽 + 𝜖𝜖𝑖𝑖𝑖𝑖
Where: 
• i indexes precincts
• c indexes clinics
• t denotes treatment assigned
• 𝛿𝛿𝑡𝑡 is a treatment fixed effect, corresponding to the assigned treatment t
• 𝛿𝛿 is a clinic fixed effect
• 𝛸𝛸𝑖𝑖𝑖𝑖 is a set of precinct specific covariates, including population, sexual assault rate, homicide 
rate, and percent rural
• 𝜖𝜖𝑖𝑖𝑖𝑖 is a regression residual
A total of 100,000 sets of possible treatment assignments (combinations of 0, 1, and 2), from all possible 
combinations, were drawn, blocking at the TCC (Kasy 2013). Using the regression specification above, 
the treatment assignment with the lowest value for the objective function was selected (Kasy 2013). The 
resulting distribution of the communities selected for the evaluation was as follows: 50 precincts selected 
for the demand-side community dialogues (Treatment 1), 50 for the supply-side service providers’ training 
(Treatment 2), and the remaining 50 to serve as control group that received neither of the interventions. 
This design allows the IE to assess the independent effects of each of these approaches for the precinct￾level outcomes. Initial power calculations conducted at the design phase (see Annex 1) indicated that the 
study would have the power to detect moderately large average program effects (0.53 to 0.65 standard 
deviations); with a recognized risk that the study would find moderate but statistically insignificant impacts 
due to the modest sample size. The ICCs and MDES values were generally found to be within the range 
of what was expected at the design phase (see Table 3). 
 
15 As previously noted, two TCCs work with fewer than three precincts, so triplicates (T1, T2, control) were not possible for 
each of the 51 TCCs.
16 The randomization technique employed was developed by Dr. Maximilian Kasy at Harvard University’s Department of 
Economics, and is based on a working paper (Kasy 2013).
8 
Table 3: Assumptions for power calculations, precinct-level outcomes
Treatment 1 & Control Treatment 2 & Control
Full Sample Reduced 
Sample
Full Sample Reduced 
Sample
Alpha 0.05 0.05 0.05 0.05
Power 80% 80% 80% 80%
Cluster level Reliability (CLR) 0.70 0.70 0.70 0.70 
Clusters (J) 71 45 74 47
Blocking variables (K) 2 2 2 2
ICC .27 .14 .25 .21
MDES 0.51 0.68 0.50 0.65
2.2.2 Individual level 
The individual-level design, involving women as the unit of analysis in Treatment 1 precincts and service 
providers as the unit of analysis in Treatment 2 precincts, was originally developed as a two-armed RCT 
in which the independent effects of each of the two programs on the proposed hypotheses could be tested
against a control group. However, further discussions with the implementer of the service provider 
training intervention, FPD, revealed that identification of a comparable control group of service providers 
would not be feasible. Accordingly, the individual-level evaluation design was modified such that only the 
demand-side intervention (Treatment 1) would be evaluated with a control group, while the supply-side 
intervention (Treatment 2) would be evaluated with a simple pre-post design.
For the demand-side community dialogue intervention, 15 women were randomly selected to participate 
in a women’s survey, one from each randomly selected household in Treatment 1 and control precincts. 
No comprehensive list of households in each precinct was available, so the data collection team worked 
with the police station in each precinct to map the precinct boundaries, and sample households within 
those boundaries using a random walk technique. Annex IV presents the random walk protocol. After 
household selection within each sampled precinct, an individual female respondent within each sampled 
household was selected for participation in the survey using a simple lottery. 
The evaluation team stratified the sample of households by two characteristics to improve the 
representativeness of the sample. First, the household sample was stratified by percentage of the precinct 
population categorized as urban, rural, or tribal, according to SAPS 2013 data. The household sample was 
drawn to be representative of the precinct on the distribution of population across urban, rural, and tribal
types.17 Second, households were stratified geographically according to the available sub-precinct 
boundaries to ensure adequate geographic coverage of the precinct. In all cases, precincts were comprised 
of several sectors, and in many cases the sectors were further delineated into subsectors. The 15 sampled 
households were distributed equally amongst the sectors and subsectors, with consideration to above 
mentioned distribution across rural, urban, or tribal group types. 
Some sectors or subsectors were excluded from the sample frame as they did not meet the eligibility 
criteria for the intervention. The intervention is geared toward women who would be likely to visit TCCs. 
The implementing partners suggested that TCCs are primarily used by women who do not have private 
 
17 For example, if 20 percent of the precinct population is categorized as urban and 80 percent as rural, three urban households 
and twelve rural households would be selected. 
9 
insurance, as a woman with private health insurance would receive private medical care. As such, the 
wealthier geographic sectors or subsectors in which the vast majority of residents would have access to 
private health insurance were excluded from the evaluation, since it would not be reasonable to expect 
that women in these households would attend a community dialogue or visit a TCC. The survey was also 
limited to those between the ages of 18 and 49. 
The power calculations indicate that the individual-level design, unlike the precinct-level design, would 
have the power to detect relatively small average program effects. Table 4 shows the assumptions for the 
individual-level design, including the Intra-class Correlation Coefficients (ICC) at the blocking level (TCC) 
and the cluster level (precinct), which were calculated from the baseline data. Based on these calculated 
ICCs, the individual-level design was estimated to have the power to detect an effect of 0.24 standard 
deviations (e.g., in knowledge of TCCs or other sexual assault centers). The ICCs and MDES values were 
found to be within the range of what was expected at the design phase. 
Table 4: Assumptions for power calculations, individual-level design
Target power: 80%
α = 5%
σ2 = 0 (fixed effects)
К = 2
J = 50
η = 15 women
Outcome ICCTCC level ICC precinct level MDES
Knowledge of the TCC .111 .132 .24
Knowledge of the TCC or other sexual assault center .086 .108 .22
Knowledge of resources for victims in community .072 .093 .22
Exposed to sexual assault awareness .057 .071 .21
Exposed to sexual assault resources .079 .084 .21
Personally know girls or women who are victims of sexual assault .030 .021 .17
2.3 DATA SOURCES AND DATA LIMITATIONS
This IE relied on administrative and governmental secondary data and primary quantitative and qualitative 
data collected by the IE team. We discuss each data source in turn: 
2.3.1 Police records
Crime statistics for each of the sampled precincts were collected from the SAPS. These data were used 
to balance the precinct sampling and assignment described above and were used as control variables during 
data analysis (i.e., in the precinct-level regressions). 
2.3.2 TCC records and supplemental intake form
In the initial evaluation design, the evaluation team proposed relying on TCC records of SGBV reporting 
and TCC utilization as the sole data sources on these outcomes. Through discussions with the NPA and 
the TCCs, SI learned that individual case records with police precinct information were not available at 
the vast majority of TCCs. Since these data are crucial for testing Hypothesis 1, the evaluation team 
developed an alternative method of capturing this data: a supplemental TCC intake form. This form was 
designed to capture basic data to track levels of SGBV reporting and TCC utilization before and after the 
10
intervention. The form is divided into two parts. Part 1 was to be filled out by TCC staff for every survivor 
when she/he first presents at the TCC. Part I contained information about the incident, the survivor, and 
planned services. Part 2 of the supplemental intake form asks about how and why the survivor decided to 
come to the TCC and is only to be filled out for survivors returning for follow-up visits at the TCC. The 
form was reviewed by the NPA and approved for distribution to TCCs to supplement existing 
recordkeeping. The supplemental TCC intake form is presented in Annex II. At endline, 47 of 50 TCCs 
submitted supplemental intake forms. 
This data collection process confronted a number of challenges. Some TCC respondents reported the 
intake form was difficult to complete, as survivors had recently experienced trauma, were tired, and had 
already answered many questions. One Site Coordinator said:
“Some of them [survivors] are not interested in having the forms filled in because they already come from two 
other offices, coming from counselling sessions and things like that. … [And] on the second visit people are 
usually … not emotionally stable because they are coming from long sessions, so … you have to explain this 
to them and then they get bored with you or irritated.”
Several TCC staff interviewed at baseline said the intake forms were very time-consuming and duplicative 
of their existing data collection processes. One Site Coordinator said, “The other day I received a new 
intake form from NPA which is going to take a lot to fill out, so it is difficult when you have somebody in 
front of you to be filling in this form then afterwards another form.” Qualitative reports suggest that other 
TCC staff were not adequately and consistently trained to use the intake forms. Furthermore, respondents 
reported that the forms were less likely to be completed when the Site Coordinator (or the trained staff 
member) was not at the TCC, especially during nights and weekends. One Site Coordinator said other 
stakeholders (such as NGO employees providing after-hours support) did not fill out the forms because 
“people don’t want to do other people’s work. ... People [are] saying I am making them do my work, so 
at this point in time I am the only person who has to fill in those forms.” Another Site Coordinator said, 
“Other stakeholders are not willing to participate because it is something that has been approved by the 
NPA and they are working for other departments, so they won’t do it.”
These types of challenges were anticipated by the evaluation team, and a number of data quality verification 
measures were conducted to encourage accurate and reliable data. SI’s data collection partner followed 
up with each TCC twice during the data collection period. The first follow-up was a phone call to each 
TCC approximately one week after the initial baseline visit. The second was a follow-up visit to each TCC 
approximately one to three months after the baseline visit to visually check for compliance and to address 
any issues or concerns. At the time of the second follow-up, 42 of the 50 participating TCCs reported 
using the supplemental intakes with at least 90 percent of survivors presenting at their TCC. Those TCCs 
with lower compliance were reminded of the data collection activities and any questions were addressed. 
Unfortunately, compliance dropped further by endline. At endline, 47 of 50 TCCs submitted supplemental 
intake forms. Of these, a lower 31 reported use of the intake forms with 90 percent to 100 percent of 
survivors. In addition, to these verification exercises, the evaluation team also compared TCC utilization 
figures provided by the NPA with intake data provided by the TCCs (see Annex IX).
Following an analysis of these different sources of data, we concluded that there were concerns with data 
from 19 TCCs and that data from another 28 TCCs was sufficiently accurate for purposes of the 
evaluation. In some cases, the data provided by the 28 cases might not be 100 percent complete; however, 
the missing data appears random and not likely to influence the outcome of the study. For example, one 
TCC collected data systematically; however, personnel did not collect data for one month, reportedly 
because they were out of paper. This gap in data collection is expected to affect all police precincts equally 
and as such should not affect the evaluation. In fact, based on the concerns identified above, we do not 
expect missing data to be correlated with police precinct or the intervention. As such, in the analysis that 
11
follows we present separate analyses of the full sample of 45 TCCs with police precinct data18 and the 28 
TCCs for which we have a higher level of confidence in the data. 
2.3.3 Women’s survey
A women’s survey was administered at households in the demand-side Treatment 1 and control group 
communities surrounding TCCs to capture community knowledge of TCCs and SGBV and attitudes 
toward SGBV at baseline and endline. Respondents were limited to women due to sample size and budget 
considerations. To allow a sufficient sample of both women and men, the sample size would have had to 
be doubled. Since the vast majority of SGBV survivors are women, the evaluation team and USAID jointly 
decided to conduct household surveys with women. The women’s survey was administered to one 
randomly selected adult woman in each of the 15 randomly selected households in Treatment 1 and 
control households using Computer Assisted Personal Interviewing (CAPI). As the respondents were 
selected based on treatment and control areas, the study is not intended to be nationally representative, 
or representative of province, and the data presented below are unweighted. Since the evaluation is 
nationwide, the survey was translated to all 12 official languages in South Africa, and it was administered 
by regionally-based teams of fieldworkers who were fluent in the languages of each area. The women’s 
survey instrument is presented in Annex II. 
There were some challenges in data collection, as the field team experienced barriers to entry in four 
police precincts near Cape Town due to high violence, crime, and drug use, coupled with racial tensions 
in those areas. To conduct data collection in this area, the evaluation team partnered with a highly localized 
data collection firm with experience working in these areas to survey households in those precincts. 
Paper-based data collection was used in these areas due to the security challenges. 
2.3.4 Service provider survey
The service provider survey measures knowledge of TCCs and knowledge and attitudes toward SGBV 
among professionals in the continuum of care for survivors of SGBV, in particular police, teachers, health 
care workers, social workers, NGO workers, and TCC staff. Respondents to the service provider survey 
were attendees of the FPD training (supply-side Treatment 2). As mentioned previously, no control or 
comparison group was possible for the service provider professionals, as it would not be possible to 
replicate the recruitment mechanism used by FPD for training participants with a group of non￾participants. The service provider survey is shown in Annex II. The evaluation was able to achieve a 94
percent response rate for trained professionals at baseline; however, each respondent did not complete 
all questions at baseline resulting in missing data, which ranges from two percent to ten percent for most 
questions (see Annex V). Of those trainees who completed a baseline survey, 75 percent were reached 
for an endline survey, for an overall response rate of 71 percent of trainees.19 
2.3.5 Qualitative TCC and NGO data
This IE utilized qualitative data collected from TCC staff and local NGOs serving survivors of SGBV to 
supplement the quantitative measures. At almost all of the 51 TCCs, an interview was conducted with the 
Site Coordinator and one NGO serving the same population as the TCC, for a total sample size of 94 
qualitative interviews at baseline. Follow-up interviews were attempted with each TCC as part of endline 
data collection. Forty-eight endline interviews were conducted with TCC senior staff, while three TCCs 
were unwilling to participate in the endline. Interview questions focused on awareness and utilization of 
 
18 This number drops from 47 to 45 TCCs because forms from two TCCs did not contain police precinct data, and are thus 
unusable for precinct-level analyses. 19 The response rate on this survey is a bit lower than hoped. The main reasons for non-response at follow-up are: respondent 
did not provide a phone number; respondent was no longer available at the phone number listed; and refusal. Up to 10 attempts 
were made to reach professionals who were simply unavailable at the time of the call. 
12 
TCCs, and TCC capacity to serve survivors. This data was used to triangulate responses from the women’s 
survey, TCC records, and supplemental TCC intake forms, and to explore alternative explanations for 
limited TCC use and TCC constraints. The qualitative data has been used to construct several TCC 
capacity indicators, which are used as control variables in regression analysis on TCC utilization. The 
interview protocols for the TCC and NGO staff interviews are presented in Annex II.
While the majority of TCC Site Coordinators and NGO staff were receptive to the interviews, several 
TCC Site Coordinators refused to participate in evaluation activities at some point during the study. First, 
the NPA representative affiliated with the USAID project relinquished her role midway through baseline 
data collection, resulting in some confusion around the study team’s permission to access TCCs. The 
subsequent point of contact for the evaluation worked to promote the project, but with varying degrees 
of success. Some TCCs stated that there was continued pressure from NPA officials not to participate in 
evaluation activities, despite NPA assurances to the contrary. 
In addition, some TCC staff were reluctant to participate in the research even when NPA support was 
clear. It is likely that some respondents did not feel they could not speak freely, despite assurances that 
the recording and notes would be kept confidential. 
Finally, one TCC was used for pilot purposes to test the TCC visit protocol and interview guide. This 
interview has been excluded from the analysis because we did not receive consent from that TCC to 
release the resulting interview notes in their original form.20 As previously mentioned, three TCCs refused 
the endline interview, citing a lack of permission from superiors to participate. 
2.3.6 Training, piloting, and data collection 
Baseline and endline data collection took place on a rolling basis over a 14-month period from August 
2014 to October 2015, in tandem with the implementing partners’ implementation schedule across all 
nine provinces in South Africa. 21 Prior to the start of data collection, fieldworkers from SI’s data collection 
partner, Impact Research International (IRI), were trained for one week on the details of administration 
of the survey and qualitative interview protocols for the TCC and NGO interviews. The training included 
definition of the roles and responsibilities for fieldworkers in various positions, sampling and all other 
relevant field protocols, research ethics, data quality assurance measures, and utilization of electronic 
devices used for CAPI. After the training was completed, both the women’s survey and the service 
provider survey were pilot tested in communities not sampled for the evaluation.22 The qualitative TCC 
and NGO interview protocols were not formally piloted as there was not an appropriate set of 
respondents not participating in the study who could serve in the pre-test. However, SI accompanied an 
IRI fieldworker to the first TCC and NGO visit to assess the adequacy of the protocols and associated 
procedures, and to make any necessary adjustments. SI evaluation team members provided oversight for 
the training, pilot testing, and the initial days of field work. Additionally, SI provided ongoing remote data 
quality monitoring for the duration of the data collection period, which included a review of sampling 
procedure for each precinct, survey data, and survey metadata on a rolling basis, and a review of interview 
transcripts and random audits of interview audio files. 
 
20 Specifically, this TCC requested that they be allowed to review and verify the transcribed notes taken after the interview. The 
resulting edited transcript appeared markedly different than the researcher’s original notes, thus the evaluation team did not feel 
comfortable including these in the dataset. 
21 Baseline data collection took place in each TCC catchment area two months or less prior to the start of the intervention and 
endline took place three months after the completion of the intervention in that TCC catchment area. 22 To pilot the service provider survey, the data collection firm gathered a group of known service providers representing each 
of the professions of interest. The study was explained to the group and they were asked to self-administer the survey. After 
completion, a debriefing focus group was held to elicit feedback and revisions were made, accordingly. 
13
FINDINGS
3.1 TCCS 
Qualitative research was utilized to contextualize quantitative data and delve more deeply into TCC 
operations and the perceptions of TCC staff. Qualitative data on TCC capacity and utilization were 
collected via in-depth, targeted interviews at baseline and endline. Initially, interviews were only planned 
with TCC staff, but at the beginning of data collection, the research team learned that many TCCs 
outsource their after-hours or specialized services to external entities, including NGOs, the Department 
of Health (DOH), and the Department of Social Development (DSD). As such, interviews were conducted 
with TCC Site Coordinators, TCC Victim Assistant Officers (VAO), and TCC case managers, as well as 
NGO management staff, nurses, clinicians, and social workers. Written informed consent was obtained 
from each participant before each interview was conducted. Respondents were presented an 
informational pamphlet about the study, asked if the interview could be recorded, and given a formal NPA 
research endorsement letter. They were then asked to review and sign the informed consent document. 
If respondents were unavailable during in-person site visits, interviews were conducted by phone. Prior 
to each phone interview, the study purpose was shared with the respondent and the consent form was 
to be signed. 
Ninety-four baseline interviews were conducted across all 51 TCC catchment areas. Interviewees included 
senior staff from 51 TCCs, representatives of 40 local NGOs—many of which were characterized as 
Victim Empowerment Programs (VEPs)—and three independent counsellors (non-NPA employed) who 
assist with survivor case management.23 Figure 4 and Figure 5 show the locations of baseline interviews 
with TCC staff, NGO staff, and counsellors. 
 
Figure 4: TCC interview sites
 
23 One TCC interview was excluded from analysis due to data quality concerns. See section 2.3.5 of this report for more detailed
discussion.
14
Figure 5: NGO interview sites
Follow-up interviews were attempted with each TCC as part of endline data collection to assess changes 
in TCC capacity, perceived changes in TCC utilization since baseline, and use of intake forms. Forty-eight 
endline interviews were conducted with TCC senior staff, while three TCCs were unwilling to participate 
in the endline. Where possible, endline interviews were conducted with the same TCC respondent 
interviewed at baseline. If the baseline respondent was unavailable for endline data collection (e.g., if the 
respondent no longer worked at the TCC), another senior TCC employee who had worked at the TCC 
at baseline was interviewed instead.
Interviews were digitally recorded and transcribed, and both voice and text data were uploaded to a 
shared and encrypted server. To ensure consistent interpretation, a team of researchers reviewed the 
qualitative data for quality control, paying particular attention to the accuracy of interview transcriptions. 
For interviews conducted in languages other than English, a native speaker translated the interview and a 
second team member fluent in that language performed an accuracy check. English transcriptions were 
used for data analysis. Coding and analysis of the transcript data was completed using Dedoose software, 
an online, cross-platform application for mixed methods research. The same two individuals completed 
coding of all transcripts to mitigate concerns of inter-rater reliability. 
3.2. TCC CAPACITY 
Factors considered in the assessment of TCC capacity included location, days and hours of operation, 
appearance and quality of facilities, services provided, staffing, recordkeeping practices, and available 
resources.
3.2.1 Days and hours of operation
At baseline, 23 TCCs were open Monday through Friday and 20 were open seven days a week. At endline, 
one TCC had extended its hours of operation to weekends and was open seven days a week. Twenty￾five TCCs provided services 24 hours a day at baseline. At endline, five more TCCs provided services 24 
hours a day, for a total of 30 TCCs. No other TCCs reported any changes to their operating schedule 
between baseline and endline. Operating hours for TCCs that were not open around the clock ranged 
from 7:00 AM to 7:00 PM, with the majority open during standard business hours (8:30 AM to 4:30 PM). 
Many TCCs that did not operate on a 24-hour schedule referred survivors to external partners for 
services after regular business hours. 
3.2.2 After-hours services
Most TCCs with on-site after-hours services received staffing support from NGOs and the DOH. Many 
had all staff members present during regular business hours, and only limited staff available after-hours, or 
had on-call staff that report to the TCC if a survivor presents after-hours. TCCs without on-site after-
15
hours services referred survivors to the casualty department of the nearest hospital for care. The hospitals 
then instruct survivors to return to the TCC during normal operating hours for follow-up, including 
counselling and other services the hospital does not provide. 
TCCs without on-site after-hours services are less equipped to provide comprehensive care, as one Site 
Coordinator described: “We [are] only operating office hours. Most of our cases were report[ed] during 
after-hours … which means victims who report to the health facility after-hours have to come back the 
following day to get the remaining services. So our biggest challenge here is that most of our victims do 
not come back.” Many respondents highlighted lack of follow-up as one of the key challenges they 
continued to face at endline.
When asked how after-hours intake and services differ from operations during regular business hours, a 
TCC Site Coordinator explained, “the process is supposed to be the same, it’s just that what was supposed 
to be done by the Site Coordinator and the VAO won’t be done the same. In this one we rely mostly on 
[NGO] staff people. … We expect them to take the victim through all the channels or all the processes 
and to ensure that the victim gets the services as expected.” The same Site Coordinator explained that 
TVP, a local NGO, also provides staff (trauma counsellors, victim advocates and general staff members), 
who perform housekeeping and cooking duties. Other TCCs reported similar arrangements, with NGOs 
providing staff support during both regular business hours and after-hours.24
3.2.3 Staffing
Even with NGO employees supplementing TCC staff, many 
TCCs reported being chronically understaffed. In 31 TCCs, 
at least one staff position was not filled at the time of the 
baseline interview, and ten other TCCs had one or more 
unfilled staff positions in the months preceding the baseline 
interview. The length of these vacancies ranged from four to 
41 months. Many respondents were unable to estimate 
vacancy lengths due to the high rate of staff turnover and 
transfers between TCC sites. Staffing shortages were 
reported across all positions: case managers, forensic nurses, 
physicians, counsellors, victim assistant officers, police 
liaisons, social workers, and support staff. Some positions were temporarily vacant due to staff 
resignations, but others reported not having the budget for full-time positions in those areas. 
Staffing shortages in all positions remained an important concern during endline interviews. Efforts to fill 
vacancies were mixed—some TCCs had hired new employees and were fully staffed at endline, while 
others still reported openings, including vacant positions of VAOs, case managers, counsellors, and nurses. 
Respondents reported that staffing shortages force existing staff members to perform dual roles or take 
on activities outside their scope of work. A Site Coordinator said she is “used to dividing myself now in 
all those roles, I have to juggle between those roles. … I try very much to be [superwoman].”
Lack of and inefficient coordination among stakeholders are two driving factors leading to long-term 
vacancies. Respondents highlighted that multiple stakeholders are responsible for filling specific positions, 
increasing the bureaucracy of the hiring process. For example, the NPA oversees the hiring of Site 
Coordinators and VAOs, DSD oversees the hiring of social workers, and DOH oversees the appointments 
of doctors and nurses. One Site Coordinator expressed frustration about her inability to fill open 
positions, saying she could not hire anyone because the TCC must take “instructions from the National 
Prosecuting Authority. Most of the things are beyond my control because we are taking instructions from 
 
24 NGO-provided services, as well as more detail about the relationship between TCCs and NGOs, are discussed in more detail 
in the “Role of NGOs as Service Providers” section of this report.
“To tell you the truth, our staff is not enough. 
... I work with six counsellors where three 
counsellors work in a day. ... They don’t get 
enough off days. It’s strenuous for them. 
Sometimes I have to go to workshops and 
sometimes the invite says I have to come with 
one or two counsellors, so it becomes a 
problem really because of the shortage of staff. 
We need people who can help us.”
-TCC Site Coordinator
16
them. The new one is from DSD and she is here because she was sent by DSD. So it’s out of my hands.” 
Another concern with the staffing structure interview respondents raised was the gender of the TCC staff 
working with survivors. While 36 Site Coordinators were women, only 16 VAOs were women. None of 
the interviewed staff were able to report that the TCC staff members had completed gender-sensitivity 
training. The majority of survivors are female and are frequently forced to interact with a male TCC 
employee, regardless of their comfort with this arrangement. Qualitative reports also suggest that space 
constraints may contribute to secondary victimization if a male TCC employee treats a female survivor in 
close quarters. A Site Coordinator explained, “The psychologist once said that we were not doing justice 
to our victims. When they do the counselling in such a small office, and sometimes we deal with people 
who were obviously raped, traumatized, and the psychologist is a man, so a lady being confined in that 
small space with another man.” 
3.2.4 Location
Many TCCs were quite difficult to locate, even for experienced research staff. Despite having coordinates 
for each TCC, the research team was unable to find 31 of 50 sites without asking for directions. In six 
cases, employees at the reception desk of the hospital within which the TCC was located were unable to 
direct the team to the TCC. At baseline, only 34 TCCs had a sign on the building/facility in which it is 
located, but by endline, all but one TCC had been signposted. This appears to have been done by design 
to prevent perpetrators from finding the TCCs; however, this is likely to severely undermine utilization.
All TCCs remained in the same physical locations at endline as baseline. 
3.2.5 Ability to meet survivor needs
At baseline, few Site Coordinators expressed high levels of confidence in their TCC’s overall capacity and 
ability to meet survivor needs. During baseline interviews, respondents were asked to rate their TCC’s 
overall capacity on a four-level scale ranging from “fully adequate” to “not at all adequate.” Only one 
respondent rated their TCC’s capacity as fully adequate, while 23 rated their TCC as mostly adequate, 19 
as somewhat adequate, and 4 as not at all adequate. One respondent who rated their TCC’s capacity as 
mostly adequate also stated that the facility was not victim-friendly. Respondents' assessment of their 
TCC’s ability to meet survivors’ needs was measured on the same adequacy scale as the overall facility 
indicator. No respondents indicated their TCC’s ability to meet survivor needs was fully adequate. 
Twenty-three said their TCC’s ability was mostly adequate, 22 said their TCC’s ability was somewhat 
adequate, and four respondents said their TCC’s ability was not at all adequate. Baseline and endline 
numbers are reported in Figure 6. 
Figure 6: Assessment of TCC ability to meet survivor need
Fourteen TCCs believed their ability to meet survivor needs had improved between baseline and endline, 
while 23 thought that their ability to meet survivor needs remained mostly unchanged from baseline. Only 
three TCCs responded that their ability to meet survivor needs had declined, which they attributed to 
staff departures and diminished NGO support. Research staff also independently assessed each TCC’s 
ability to meet survivor needs at endline on a four-part scale from “fully adequate” to “not at all adequate” 
2
27.1
46
43.8
38
29.2
8 6
0% 20% 40% 60% 80% 100%
Baseline (%)
Endline (%) Fully adequate
Mostly adequate
Somewhat adequate
Not at all adequate
Unsure/mixed
17
Thirteen TCCs were rated fully adequate, 21 were mostly adequate, and 14 were somewhat adequate, 
which represent a notable improvement from baseline. Overall, improvements in ability to meet survivor 
needs and TCC capacity were reported by both TCC staff and the data collection team.
3.2.6 Communication and collaboration with stakeholders
Many TCCs at endline reported increased communication and collaboration with stakeholders, including 
the NPA, SAPS, DOH, NGOs, and hospitals, as a factor that improved their TCC’s ability to meet survivor 
needs. Nearly all respondents reported strengthened relationships and more frequent communication 
with stakeholders. At baseline, many TCCs reported holding regular implementation meetings during 
which stakeholders come together to discuss TCC issues and challenges. At endline, these meetings were 
repeatedly cited as one of the key factors that contributed to improved TCC service delivery. One Site 
Coordinator said, “wherever there are problems, we try and actually discuss and actually come out with 
resolutions as to what we need to do to move forward, so most of the stakeholders are aware of what 
things are happening at Thuthuzela.”
One TCC created a “cluster reporting template whereby each and every stakeholder will actually report 
based on that particular template and we will take it as our report card to our monthly and implementation 
meetings.” No other respondents reported using such a template, but most identified increased 
communication and collaboration with stakeholders as a catalyst for better survivor care. One Site 
Coordinator explained:
“Last week when we were doing our year plan, we were emphasizing that we do things together, let’s 
invite each other when we have campaigns, when we have issues. When one has a challenge, let’s liaise 
so that the very same victim gets necessary services. … [If] we don’t work together the same victim might 
not get the very same service.”
Another Site Coordinator described how they overcome challenges with stakeholders and balance 
competing organizational interests: 
“We [are] bound to clash when we have different norms and values from our different departments. But 
at same time we remind ourselves that as much as my values, my objectives are from my department are 
same, convert and whatever and use to give medical care but at the end of the day we still have to provide 
good service to the victims. Let’s combine that and give them good service.”
In addition, increased Internet connectivity at TCCs facilitated stakeholder engagement since this allowed 
them to communicate with stakeholders by email.
3.2.7 Space constraints
Some TCCs reported deficient facilities as a barrier to fully meeting survivor needs, and insufficient space 
was repeatedly cited as a significant problem. Only four TCCs rated their space as fully adequate, 20 as 
mostly adequate, 14 as somewhat adequate, and 12 as not at all adequate. At endline, research staff 
independently assessed each TCC facility’s space, finding 16 TCCs to have fully adequate space, 20 as 
mostly adequate, eight as somewhat adequate, and four as not at all adequate, as outlined in Figure 7. 
8
33.3
40
41.7
28
16.7
24
8.3
0% 50% 100%
Baseline (%)
Endline (%) Fully adequate
Mostly adequate
Somewhat adequate
Not at all adequate
18
Figure 7: Independent assessment of TCC space
While the research staff’s assessment indicates that improvements had been made by endline, but 
respondents underlined the numerous problems inadequate office space creates. For example, when 
police bring suspected perpetrators to the TCC for forensic examination and evidence collection, 
sometimes survivors and suspected perpetrators must wait in the reception area at the same time, as a 
Site Coordinator explained: 
“We have only one examination room and we still get perpetrators that come in for DNA, and we don’t 
have a suspect room whereby the suspect can be examined. So it happens that the suspect will also use 
the same room as the examination room that is also used by survivors, and because of that you will find 
that sometimes the suspect will be coming for DNA while there is a victim coming for rape.” 
Another Site Coordinator echoed this concern: 
“I feel that our clients don’t feel comfortable sitting in our waiting room because they were using the same 
door [as suspected perpetrators]. We have only one entrance. So imagine you were raped and you see 
the police coming in… and you as the victim were trying to digest what happened to you and seeing that 
up and down happening in the waiting room.” 
Space constraints lead to administrative offices being used for multiple purposes, including as examination 
or consultation rooms. One Site Coordinator explained that their TCC was not built according to NPA 
specifications, which has caused privacy concerns: 
“The Thuthuzelas are built according to a certain blueprint and this one that we have here is not built 
according to the Thuthuzela blueprint. So we have been trying to negotiate to have a Thuthuzela built for 
us here. You see, when a victim comes to the reception, this is the counselling room and there is always a 
clinic as you can see the chairs outside there. … There is no privacy.”
At endline, a TCC Site Coordinator explained that they were still operating out of a temporary office 
space:
“We are working in a park home, I believe this is a temporary measure, definitely we would want a 
permanent structure maybe with rooms where we can have a rest room for our victims while they are 
waiting for assistance from the sisters and the doctors. … it becomes very hot in here even when we have 
the [air conditioning], therefore if we have a permanent structure we can be able to engage with the 
hospital to come and install maybe a smaller dispensary area whereby we know that when our victims 
comes in they don’t have to walk to the hospital pharmacy for medication, but can have all the medication 
they need here and everything is done here, rather than having them walk to the hospital and maybe see 
the perpetrator and experience secondary victimization.”
3.2.8 Facility appearance and supplies
The condition and upkeep of facilities is also a source of concern. At baseline and endline, research staff 
independently ranked the appearance of each TCC on a four-level scale ranging from “very nice” to 
“poor,” as shown in Figure 8. At baseline, 12 TCCs were assessed as very nice, 15 as somewhat nice, 12 
were in some disrepair, and eight were in poor condition. However, the overall appearance and upkeep 
of the facilities considerably improved by endline, with the majority (42) rated as very nice or somewhat 
nice. Only two remained in some disrepair and three were in poor condition at endline. 
19
Figure 8: Independent assessment of TCC appearance
A Site Coordinator of a TCC located within a hospital reported that the hospital rarely extends its cleaning 
or maintenance services to the TCC, even intentionally omitting the grass in front of the TCC when 
mowing. The Site Coordinator stated that “visitors of the hospital will throw cans of alcohol in front of 
the TCC because they were seeing that it is dirty.” The issues were repeatedly reported to the NPA and 
the DOH, but they did not take steps to address the matter. TCC staff reported cleaning the office 
themselves and paying for maintenance and mowing services with their own money because TCCs do not 
have the budget for these services. 
In addition to space limitations and poor quality facilities, 84 percent of TCCs were deemed to have 
inadequate connectivity (both phone service and Internet connections) at baseline. Some TCCs are not 
equipped with landline phone service, so staff must rely on personal cell phones to make calls. TCCs that 
have landlines have limited budgets for phone service. One respondent explained:
“We were allocated R100 a month for calls, which is not enough to communicate with survivors to remind 
them of their next appointment … and to call standby staff. … We end up using our own money for 
work-related calls which is not paid back to us.” 
Staff also reported having to go to nearby Internet cafés to use the Internet at their own expense. By 
endline, many TCCs had Internet access or had plans to install it in the near future. Some TCCs had 
increased phone service, but others were still using personal cell phones. By endline, 27 TCCs reported 
receiving new resources (e.g., computers, fax machines, photocopiers, office supplies, etc.) while the 
remaining 21 TCCs reported that they had not received any new resources since baseline. The research 
team independently assessed the overall adequacy of each TCC’s supplies at endline, the results of which 
are compared in Figure 9: 22 were fully adequate, 18 were mostly adequate, and eight were somewhat 
adequate, an improvement from baseline.
Figure 9: Independent assessment of TCC supplies
3.2.9 Medical supplies, clothing, and food
Most TCCs were adequately stocked with medical supplies, which are provided by DOH and the hospitals. 
Two TCCs did not have colposcopes (a specialized medical camera used to photograph genital injuries 
25.5
42.6
31.9
22
25.5
4.3
17.1
6.4
0% 20% 40% 60% 80% 100%
Baseline (%)
Endline (%) Very nice
Somewhat nice
Some disrepair
Poor condition
16
45.8
38
37.5
32
16.7
14
0% 20% 40% 60% 80% 100%
Baseline (%)
Endline (%) Fully adequate
Mostly adequate
Somewhat adequate
Not at all adequate
20
invisible to the unaided eye) at baseline, but one of these TCCs had received a colposcope by endline. 
Nearly all TCCs reported having sufficient supplies of comfort packs, which include personal hygiene items 
and underwear. Comfort packs are typically provided by DOH, NPA, and NGO (and occasionally 
corporate donors such as Nivea and Colgate, which 
were reported at endline), but they do not generally 
include clothing other than underwear. Many 
survivors come to TCCs with dirty or bloody 
clothing, which must be collected as evidence if they 
choose to file a police report, yet most TCCs do not 
have a sufficient supply of clothing to offer survivors. 
NPA and DOH provide tracksuits to a few TCCs, 
but some staff members reported purchasing 
clothing themselves. Other sites rely on clothing donations, but one TCC Site Coordinator explained that 
TCCs are not allowed to solicit donations because they are government-operated. Individuals may make 
voluntary clothing donations to the TCCs, but donors must write letters explaining why they want to 
donate the clothes, as well as where and how they acquired them, even for second-hand clothing, which 
could present an impediment to would-be donors.
Food supplies are also largely inadequate. The NPA previously supplied groceries to some TCCs, but it 
has largely ceased that practice. One TCC employee said that RTI International used to provide their 
groceries, but that support has also been discontinued:
“Ever since we came into the NPA, we haven’t received anything, and those used to assist especially with 
these long waiting hours, then you know you can give them tea, and now it’s cold, give them tea, give them 
something to warm themselves up. … So right now [a local NGO] is buying groceries, but sometimes that 
is not enough. It gets finished before the time comes for them to buy, then we have to wait for them to 
buy because the NPA does not give groceries at all.”
Many TCCs that do provide food service utilize hospital kitchens to prepare it. However, that option is 
not readily available outside of regular business hours or to TCCs that are not located in hospitals. Several 
TCC staff members reported purchasing food for survivors with their own money.
Survivors who first report to a police station and those who travelled great distances to reach a TCC may 
not have eaten for extended periods of time, so providing food to survivors at TCCs is considered a 
critical need. TCC awareness campaigns advise survivors not to eat before medical examinations, which 
can further extend the length of time that a survivor has not eaten. A TCC employee illustrated how long 
a survivor could go without food: “Let’s say that the victim was raped today and could not find transport 
to come to the TCC and only comes the following day. Remember you told the victim not to eat until 
the medical examination is done. Look at that, they won’t get anything when it comes to food.” 
The need for food can also be related to treatment, as PEP medications should be taken on a full stomach. 
PEP must be taken within 72 hours of a sexual assault to reduce the likelihood of HIV transmission, but 
can have negative side effects if taken on an empty stomach. 
3.2.10 Transportation
Survivors’ difficulties in getting to the TCCs and a lack of funding for transportation were repeatedly cited 
as one of the TCCs’ most pressing challenges. If a survivor does not have access to transportation, s/he 
may be unable to come to a TCC for an initial visit or may not receive continuity of care during follow￾up visits. Most TCCs do not have a TCC-owned vehicle or funding to reimburse survivors for 
transportation costs. A TCC Site Coordinator explained: 
“The stats on our follow-ups were very alarming because transport as a resource is a problem. To get 
people to come here for follow-ups, they have to travel, and sadly a lot of our clients were unemployed. 
They come from rural areas and they don’t have the means to come to town. Some of them even walk.” 
“The challenge now is clothing. You might find that a 
victim came and was beaten or had been stabbed. 
He/she is bleeding and the clothes they were wearing is 
bloody, so we don’t have clothing. Even though we have 
clothes, it is not enough. Even when you give the victim, 
you don’t give them decent clothes, you give them 
something shabby.”
-NGO Trauma Counsellor 
21
Certain transportation costs are reimbursed, but generally only for survivors traveling to and from court 
to testify or attend legal proceedings. TCCs do not 
have the budget to reimburse survivors for follow￾up visit transportation expenses. Failure to return 
for follow-up can have serious health implications, 
as survivors may not take the full course of PEP. 
Most survivors receive a PEP starter pack during 
their initial visit and are required to return to 
receive the remainder of the PEP medication. Some 
TCC staff reported giving survivors the full course 
of PEP if they know they will not be able to return to the TCC, although this practice is not common. 
One Site Coordinator explained this as follows:
“Sometimes a person will come only for the first visit and don’t do their follow ups. We still have that 
challenge but when it comes to the initial visit we try to help them. Actually the policy says we must give 
them prophylaxis for three days or five days then when they come back and finish the course so we 
decided not to give it to them because if we give them for three to five days and they don’t come back it 
means they will default and it might happen that they are infected. Now we have a strategy to give them 
for the whole 28 days on the initial visit.”
Lack of transportation also inhibits the TCCs’ ability to conduct outreach activities and awareness 
campaigns around the community. 
3.2.11 Funders and supporters 
At endline, respondents were asked if their TCC had any new funders or supporters. Very few TCCs 
reported any changes in their sources of funding or support. Some respondents did not know details 
about the sources of their funding, but most respondents who were able to describe their funding 
structure identified South African government agencies (e.g., DOH, DSD, NPA) as their main funders. 
Several TCCs cited funding and/or support from local NGOs, international organizations (e.g., UNICEF), 
and foreign governments (e.g., the Danish Embassy). 
3.3 TCC UTILIZATION
3.3.1 Survivor demographics 
The age and sex of survivors utilizing TCC services varies significantly between TCCs and even within 
individual TCCs over time. Some TCCs report that children ages 0-12 make up the largest group of the 
survivors they serve, with some estimating youth under the age of 18 comprised as much as 80 percent
of all survivors. The other most common survivor demographic is females over the age of 18. 
Adult males are the least frequent users of TCCs. One TCC said they would sometimes serve one or 
two adult males a month, but it was common to not serve any adult male survivors for months at a time. 
Many adult male survivors are prisoners brought in by the Department of Correctional Services. There is 
a significant stigma associated with male victimization that likely contributes to the rarity of reporting. A 
TCC doctor explained that adult males:
“don’t want to go and expose themselves, because they have a problem and then go to a police station, 
and you go there you see all sorts of sexual assaults, police officers mock them, and that is where it’s very 
demoralising, and that is why we find that males do not come for help, unless the family knows about it 
and they bring them. But to go to a police station now and say that I was abused, it’s very demoralising to 
the man, I think the police officers need to be more sensitive to this issue because it’s a problem in our 
society.” 
Most respondents said that lesbian, gay, bisexual, transgender, and intersex (LGBTI) individuals rarely 
reported to TCCs for care. However, several respondents noted that TCCs do not ask about a person’s 
“[Survivors] have to come for follow-ups and their problem 
would be that they are not working and they don’t have 
money. We are also unable to go to them. Other TCCs have 
NGOs that give them money so that they come back for 
follow-ups. Here we don’t have that, and we are unable to 
go them because we don’t have transport, we also don’t 
have resources to get to them.”
-TCC Site Coordinator
22
sexual orientation during intake or at any point during the consultation, so unless a survivor self-identifies 
as LGBTI, there is no recorded data for this demographic group. It is likely underreported due to social 
and cultural stigma of LGBTI identities. One TCC Site Coordinator said that sexual violence against LGBTI 
people is “about correction. They [perpetrators] think they can correct people to be straight. This one 
case they burned inside, he ran away naked in the street. It was traumatic.”
At endline, nearly all TCCs reported that the volume of survivors presenting at the TCCs for care was 
either similar to baseline levels or slightly higher. Many respondents mentioned that the number of 
survivors reporting to TCCs fluctuate by season, and is higher around the time of various events when 
people may be more vulnerable (e.g., festivals). A Site Coordinator explained the differences in 
vulnerability between child and adult survivors and how vulnerability changes throughout the year:
“I think in winter they were not vulnerable. They were indoors as it is cold. In summer they were outdoors, 
partying and that is where they were vulnerable. If the weekend is rainy and cold, we were happy because 
then it is quiet. We will say at least people were not raped. It is nice. With children, they were raped inside 
homes, so yes those were the types of cases we usually get, the serious cases that we get in winter. These 
were the types of cases were they break down your door and they come into your house.”
Some TCCs that reported increases in the number of survivors reporting to their facility credited 
outreach activities for increasing awareness of TCCs and their services: “Sometimes during events, you 
see massive public awareness campaigns on the media, like 16 days of activism and so forth, that is when 
you find people coming forward to report, not knowing if this is created by the media hype or what, or 
the reasons are the awareness campaigns.” A few respondents attributed outreach activities as the cause 
of reduction in overall prevalence of SGBV in their communities. 
3.3.2 Referrals to TCCs
Survivors are referred to TCCs from a variety of sources, including police, hospitals, clinics, schools, 
NGOs, social workers, and churches. Some TCCs cited schools (teachers, Department of Education social 
workers, and school social workers) as their primary referral source for cases of child sexual abuse. A 
TCC Site Coordinator cautioned that teachers were not always knowledgeable about TCC referral 
procedures, and “sometimes you will find that they do not take it seriously. There is this thing with our 
people, if they don’t see an injury and if the child walks normally then the child can’t have been raped.” 
Some respondents indicated that teachers might be hesitant to report cases because they are afraid of 
getting involved in the legal process: 
“Let’s say a child has disclosed to the caregiver at school or a caregiver has identified a case of an abused 
child and that child speaks out, you find that that caregiver is afraid to report the case, they are afraid to 
be the first reporter because they are afraid to go to court ... they would say that there is such and such 
a case but I want to be anonymous.”
Another respondent said primary healthcare clinics have a poor understanding of where to refer suspected 
victims of sexual violence because:
“they are not part of the [implementation] meetings, so they do not know where to refer exactly, that is 
why you will find that some of them will refer just to the hospital, … we need to also bring them in as 
part of the stakeholders, as much as they see our patients if they come across such problems they need 
to be able to know where to refer and what is it that they should not do.” 
At endline, TCC respondents described targeting awareness campaigns and outreach activities to potential 
referral sources, like schools and clinics. These activities were designed to teach stakeholders to identify 
potential SGBV survivors and know when to refer them to the TCC. Respondents underscored the need 
for continued stakeholder engagement and education to address these misperceptions and gaps in 
knowledge, and indicated that their efforts have been successful so far: “Other stakeholders now have a 
better understanding of the TCC model because we do presentations, meetings to explain the importance 
23
of referring victims to the TCC. We have better communication channels and everybody seems to be on 
board.”
The police are another significant referral source, with one TCC estimating approximately 90 percent of 
its cases are brought in by police. TCCs work extensively with SAPS, because the police are usually 
involved at some point throughout the process, even if the survivor was not brought in or referred by 
them. Survivors who first report their experiences to the police are subsequently brought to a TCC for 
examination. If survivors first present to the TCC, the police are usually called, but survivors are not 
required to make a police report before receiving care from the TCC. Most respondents reported positive 
working relationships with police stations in their area, but some expressed frustration with police officers 
that were not familiar with the referral procedures. 
One Victim Assistant Officer said some police stations “don’t know where the TCC is, they don’t come 
with the correct forms. Sometimes they bring victims in a van or the victim will sit in between two officers 
at the front. Sometimes they take longer to come, and at times they don’t have cars. Some don’t bring 
the correct crime kits.” Respondents observed that the police’s treatment of survivors does not always 
meet TCC standards of support and compassion. Some respondents described problems with how 
survivors were transported to TCCs, particularly if they were forced to ride in the back of the police car 
or between two officers in the front. Many respondents reported that survivors endure long wait times 
at police stations without the opportunity to clean up or change clothes before being brought to the 
TCCs, which can be embarrassing and secondarily traumatic. 
By endline, 67 percent of TCCs reported that police precincts were bringing or referring more survivors 
for treatment. Some respondents identified higher crime rates in certain precincts as the reason for the 
increase, but others pointed to the increased collaboration and outreach activities:
“For the past few months the police service had a lot of awareness-raising campaigns, but also clinics, and 
even the FPD and Soul City together with us, we also had our awareness-raising campaigns these past few 
months, [which] make people aware that there is such a service like the TCC."
3.3.3 Barriers to accessing TCCs
Respondents identified numerous factors that pose 
barriers to access and utilization of TCCs. TCC 
location and access to transportation are two barriers 
that were discussed in more detail in section. 
Respondents also noted a lack of knowledge about 
TCCs more generally—many people are not aware 
that TCCs exists or have misperceptions of what 
services TCCs provide. A VAO explained that some 
hospitals refer people who have not experienced 
sexual violence to TCCs for general counseling or 
social work services. A TCC Site Coordinator said, 
“Even in the hospital itself, some of the nurses don’t know what is being done here, so we try our best to 
engage them and to do training.” 
This problem is demonstrative of a broader misunderstanding of the scope and purpose of TCCs, with 
many people believing TCCs provide shelter and services to individuals experiencing homelessness or to 
people in crisis situations. There is also a misperception that TCCs do not treat male rape survivors, only 
providing care to women and children, which is untrue. Others reported thinking that people who go to 
TCCs would be arrested for other crimes. Some people thought that TCCs provide assistance with 
applications for South African Social Security Agency (SASSA) grants and offer general social work services. 
Others believe that going to a TCC guarantees a conviction in court, which was described as a source of 
frustration for survivors whose attackers were not convicted.
“They think we offer shelter for homeless people and 
all those things, even some of the police members 
don’t understand. Even the guys from the emergency 
medical services don’t understand, they will just get an 
old lady who is staying on the streets and then they 
bring them to the hospital and they say no, just take 
her to the TCC and when you interview them and you 
don’t find anything related to a sexual offence, you 
must say there is nothing we can do unfortunately.” 
-TCC Site Coordinator
24
TCCs have attempted to combat the lack of information and misinformation with workshops, outreach 
campaigns, and radio broadcasts. A TCC Counsellor explained, “we do awareness campaigns in schools 
and everywhere, to teach people that if something like this happens to you, you don’t have to hide in the 
house, you must come to the TCC where you will get help and support. Last month we had an awareness 
campaign at two schools, and we also go to our local radio station to talk about the services the TCC 
provides.” These awareness campaigns were designed to increase knowledge of the existence of TCC 
facilities and the services they provide, as well as dispel stigma surrounding sexual violence and address 
the roots of broader sociocultural factors that contribute to SGBV. 
Stigma and fear were two of the most cited barriers to TCC utilization. An NGO coordinator said, “some 
people still think that if you were raped you asked for it, because maybe you wore short clothes or you 
drank someone’s booze now you have to pay.” A TCC Site Coordinator also said, “people were still 
scared to report rape because of stigma in the society. People don’t want to be associated with being 
raped. People were scared of the perpetrators.” The fear of retaliation also presents a barrier and can be 
especially challenging for survivors who live near their attackers because they must continue to interact 
with them after the attacks. Survivors also fear not being believed. A Site Coordinator explained how 
these two fears are related:
“We found that a lot of our survivors, the perpetrators were related to them or they were their husbands, 
fathers, stepfathers of girls, so it is that fear of not being believed, it is the fear of the family, intimidation 
and those types of things. … They don’t want to go to the police because they fear they will be laughed 
at or maybe they had a few drinks and now they feel guilty. We get those cases where they go to the 
police and they’ve had a few beers and the police tell them ‘no come back tomorrow, ’go sleep off that 
beer or whatever,’ and then they don’t feel like returning the next morning.” 
Many respondents described a cultural norm prescribing sexual violence be kept within the family and not 
reported to the authorities. There were also widespread misperceptions that a survivor is required to file 
a police report or bring charges against a perpetrator to utilize TCCs. In reality, TCCs provide medical 
and counseling services to any survivor, regardless if s/he chooses to pursue criminal action. A Victim 
Assistant Officer explained the process TCCs follow in this circumstance: 
“If a victim does not want to go to the police and they know about the TCC and they come straight to the 
TCC, we will assist the victim, but we will also inform the victim that they can lay a charge. If they then 
want to lay a charge, we will call the police or the detective to come out. But we will not refuse services, 
even if the victim indicates they only wants a medical examination and PEP, we will assist that victim, even 
when she changes her mind later we will assist that victim. We will not refuse a victim from services at 
the TCC.” 
A TCC Site Coordinator added, “We do take the evidence so that if they change their minds, and they 
want to open a case later, they can have the evidence. We keep the specimen for six months, then after 
that if they don’t come back we discard it.”
3.3.4 NGOs as service providers
Given the limited hours of operations, financial constraints, and the numerous and diverse needs of 
survivors, TCCs depend heavily on NGOs for support. TCCs partner with many NGOs that extend and 
integrate services throughout the community. Representative organizations include Mosaic, an NGO that 
serves domestic violence survivors, OPTIONS, an NGO that works with crisis pregnancies and HIV, and 
the Association of Persons with Disabilities (APD), an NGO that works with people with intellectual 
disabilities, among many others. Key areas of NGO-provided assistance are after-hours support, 
psychosocial counselling, and outreach activities. Some NGOs provide temporary housing or shelter to 
survivors. LifeLine and ChildLine are two NGOs with which TCCs frequently partner. A counsellor said 
LifeLine “helps the broken souls, abused children, and women. … We are dealing with abuse, marriages, 
divorces and different types of counseling, trauma and debriefing.” ChildLine provides similar services, but 
25
for youths ages 17 and under. A counsellor said ChildLine counsels survivors “so that they were able to 
cope with the situation that they find themselves in. We also make sure that we empower them so that 
they were able to go on or to cope with the traumatic situation which they have been exposed to.” The 
same counsellor highlighted ChildLine’s education initiatives and outreach campaigns in schools and 
throughout the community to educate children about sexual 
violence. Families South Africa (FAMSA) is another NGO that 
provides trauma counselling to survivors of sexual violence and 
domestic violence, as well as marriage counseling. 
Nearly all NGO representatives reported favorable, productive 
working relationships with TCCs. A LifeLine site manager 
explained that despite occasional interpersonal disputes, patient 
care is always prioritized: 
“We work well together, you know that complications were here and there, but when it comes to the 
victim, we put aside our differences because the victim comes first. We work together. If we have our own 
vendetta, we will solve them later. But the good thing is that if there is something I don’t like that 
Thuthuzela is doing, I confront her [the Site Coordinator].” 
Most NGO respondents reported similar attitudes. A Lifeline Site Manager said that the TCC she works 
with uses their monthly implementation meetings (also discussed in section 3.2.6) to air any grievances 
and work collaboratively to find solutions to problem. These meetings facilitate dialogue and give NGO 
staff a better understanding of TCC constraints. A lack of resources, such as telephones and 
transportation, can negatively affect patient care, but coordination with external stakeholders, like NGOs, 
can help alleviate these problems. An NGO counsellor commended TCC employees for their hard work 
despite their limitations:
“They perform well and beyond their mandates, they sacrifice their time to help the survivors. … They 
work tirelessly to provide the best service and ensure that the clients were served with respect. They 
perform well under those circumstances. They don’t complain that they don’t have all the resources they 
need. They do their work, improvise where necessary, ask for help elsewhere to get the job done.”
Coordination with NGOs and other stakeholders fills service gaps that TCCs are not always able to 
address and improves the overall quality of care given to survivors. A LifeLine social worker described 
how LifeLine and other stakeholders backstop the TCC’s efforts: 
“Let’s say our patient needs a home. We at LifeLine will get a home but the problem would be transport, 
so we will help each other again—maybe SAPS will provide transport so that we meet the patients’ needs, 
because at the end they have to be holistic. They cannot as a TCC provide all the services, maybe medically, 
but socially if the patient does not have a place to stay, it’s not holistic. So we need to provide the patient 
holistically in every way and be able to provide everything they need.”
Even with TCC work being supplemented by NGO efforts, TCCs continue to face challenges in providing 
holistic care. One TCC Site Coordinator acknowledged the TCC’s shortcomings and identified
empowerment and vocational training as key areas for growth. Vocational training is especially important 
for women who were financially dependent on their abuser. The Site Coordinator said: 
“We need to do empowerment, so in this area we still lack. Maybe if we have like an NGO where once 
they were done with their counselling we give them business management skills and teach them how to 
start their own businesses, teach them anything that they can use to put food on the table, because that’s 
when we see them dropping cases and getting confused. To say at least this person even though he raped 
my child we never slept with an empty stomach, that’s when you get all of those issues and once you come 
with R5000, to say please withdraw the case you ask yourself twice, I am hungry, there is money, I am 
the one who went there to open the case I can still go there and withdraw the case and some even were 
“The TCC is providing a remarkable 
service to its victims; people regain their 
dignity and respect after they have been 
to the TCC, especially those that complete 
their follow-up sessions. Their mission 
statement says that they turn victims into 
survivors and that is exactly what they do.”
-NGO Social Worker
26
coached to say that the person just disappeared so that the police cannot find them, so the case will be 
withdrawn then they take the money. Empowerment for me I think is the key.”
3.4 IMPACT OF INTERVENTIONS ON TCC UTILIZATION 
3.4.1 Description of supplemental intake data submitted by TCCs
As part of the data collection for this evaluation, TCC staff were asked to complete a supplemental intake 
form for each survivor presenting to the TCC between the baseline and endline TCC visits. Of the 51 
TCCs, 47 submitted supplemental intake data to the evaluation team. According to these records, TCCs 
together served 3,125 survivors, an average of 74 survivors per TCC, over the approximately five-month 
evaluation period. The range varied dramatically with six TCCs reporting an aggregate of fewer than 10 
cases over the 5-month period and 10 TCCs reporting over 100 cases. Amongst this full sample of TCCs, 
this is a median of 10.6 cases per month. As the rate of TCC use was lower than anticipated, the evaluation 
team undertook a series of data verification analyses, detailed in section 2.3.2, which included historical 
data from some TCCs. These historical data corroborate the levels of use measured through this IE for 
those TCCs we have kept in our sample. 
Based on data verification analyses, the data from 19 of these TCCs was deemed to be less reliable as 
these TCCs were believed to have completed intake forms for fewer than 70 percent of the survivors
presenting at their TCC. For the descriptive analysis, we restricted the sample of remaining 28 TCCs 
further, to 24 TCCs that are estimated to have completed the supplemental intake forms for at least 90 
percent to not present underestimates in TCC utilization. The larger sample of 28 TCCs was used in the 
regression analysis for reasons explained in section 2.3.2 of this report. 
Among the reduced sample of 24 TCCs that are believed to have kept accurate intake records, intake of 
survivors is still fairly low, with most TCCs receiving fewer than 20 survivors per month with a median of 
12.7 cases, averaged across all months between the TCC baseline and endline visits (see Figure 10). As is 
also evident from Figure 10, the caseload varied substantially from just 4 survivors per month in one TCC
to 48 survivors in another. It should be noted that NPA data indicate higher monthly caseloads, with a 
median of 22.5 cases per month (see Annex IX). 
Figure 10: Average number of survivors presenting to each TCC per month (n=24)
0
5
10
15
20
25
30
35
40
45
50
TCC
1
TCC
2
TCC
3
TCC
4
TCC
5
TCC
6
TCC
7
TCC
8
TCC
9
TCC
10
TCC
11
TCC
12
TCC
13
TCC
14
TCC
15
TCC
16
TCC
17
TCC
18
TCC
19
TCC
20
TCC
21
TCC
22
TCC
23
TCC
24
Monthly cases
27
3.4.2 Balance statistics at baseline 
First, we examine whether the Treatment 1, Treatment 2, and the control samples are comparable on 
key variables. As explained in section 2.2.1, precinct-level (cluster-level) sampling and random assignment 
was designed to balance the treatment and control groups on four variables of interest: population size, 
sexual assault rate, homicide rate, and percentage of the population that is rural. As such, the Treatment 
1, Treatment 2, and control precinct sample sub-groups are balanced on these four characteristics. Given 
that data quality concerns required us to limit the sample in some of the models in the analysis to a subset 
of precincts, there was a risk that the final samples of treatment and control precincts would be 
unbalanced. This risk was mitigated by the design, which clustered treatment and control precincts around 
TCCs. If a TCC is excluded from the analysis sample, an equal number of treatment and control precincts
would thus be excluded from the analysis. As such, the balance statistics for the reduced sample shown in 
Table 5 also show balance on all key variables. 
Ideally, the sample would also have been balanced on the number of survivors presenting to the TCC in 
each precinct at baseline. However, this information was unknown before this evaluation; thus, it was not 
possible to balance on this important characteristic. Fortunately, balance statistics for both the full sample 
and the reduced sample confirm that our treatment and control groups are statistically similar on this key 
variable. Results from independent sample t-tests are displayed in Table 5 and show that, as was the case 
of our original sampled precincts, the reduced sample of Treatment 1, Treatment 2, and control precincts 
is balanced on all key variables. 
Table 5: Balance statistics for the reduced sample at baseline
Treatment 1 (n=44) Treatment 2 (n=47)
Control 
(%)
Treatment 
(%) p-value Control 
(%)
Treatment 
(%) p-value
Survivors per month 3.3 4.6 .230 3.9 3.1 .573
Population 66,467 86,717 .147 66,447 73,993 .581
Sex crimes 151.8 153.0 .958 151.8 151.8 .999
Murders per 100K 35.6 37.0 .837 35.6 30.6 .475
Rural (yes) .09 .06 .582 .09 .08 .831
3.4.3 Impact of community dialogues on TCC utilization
Multiple regression analysis using the difference-in-difference (DiD) method was conducted to test 
whether the community dialogues (Treatment 1) led to increased use of TCCs, as measured by the 
average number of survivors per month presenting to the TCC before and after the intervention in 
Treatment 1 and control precincts. Since the distribution of TCC utilization was skewed and due to a 
curvilinear relationship with some variables, the log of the average number of survivors treated per month
and several control variables was used in the analysis presented in Table 6. 
The analysis also controls for the rate of sexual assault in the precinct, population size, and the murder 
rate, as another measure of crime prevalence. In addition to these, the qualitative observational data 
collected at TCCs was used to construct several indicators of TCC capacity to include in the analysis. 
These comprised a facility assessment, supplies assessment, space assessment, and overall assessment of 
the ability to meet the needs of survivors. SI’s local researcher scored each of the TCCs on a scale from 
1 (not adequate) to 4 (fully adequate) based on observational data from the qualitative interview 
28
transcripts.25 Also included was a measure of whether the TCC reported having one or more positions 
unfilled at the time of the baseline interview and if the TCC operates 24 hours a day every day. 
Four different model specifications are presented in Table 6. Models 1 and 3 use the full sample of 45 
TCCs that submitted data to the evaluation team with control variables. Models 2 and 4 only include those 
28 TCCs for which we have greater confidence in the reported utilization. The first and second models 
present the full and reduced samples, respectively, but exclude control variables. Since we randomly 
assigned precincts to treatment and control groups and our groups balance on these variables, control
variables are not technically necessary. In models 3 and 4 we include control variables to better understand 
what factors do explain utilization and to increase our power. Regardless, as we would expect, the 
treatment effect estimates under models 1 and 2 are very similar to those in models 3 and 4. 
The DiD treatment effect variable in Table 6 provides the test of our Treatment 1 hypothesis. This variable
shows no impact of the community dialogues on TCC use across all four models. Because the coefficients 
in Table 6 are based on the logged average number of survivors per month, Figure 11 translates these 
regression coefficients to real effect sizes in an easy to interpret format. At baseline using the full sample, 
control precincts had 3.3 cases per month on average while treatment precincts had 3.9 cases. At endline, 
the control precincts had 2.7 cases per month on average and treatment precincts had 3.5 cases, an overall 
drop of 0.2 cases per month in Treatment 1 precincts relative to control precincts from baseline to 
endline. The observed general time trend of declining cases from baseline to endline in the full sample is 
likely attributable to declining compliance on the part of TCCs in using supplemental intake forms rather 
than a true decline in TCC use and should thus be interpreted as such. Nonetheless, as previously stated, 
this should not differentially affect our treatment or control precincts and therefore the estimated 
treatment effect.
Figure 11: Difference-in-difference estimate of the effect of community dialogues on TCC use
Several control variables were statistically significantly associated with higher TCC use. Population size of 
the police precinct was the strongest driver of TCC use across both model 3 and model 4 (p<.001). The 
official rate of reported sex crimes was a statistically significant predictor of TCC use in model 3 (p=.068), 
but not in model 4. The murder rate, which is included as a proxy measure for actual sexual assault rates 
given underreporting, is also highly statistically significant in both model 3 (p=.014) and model 4 (p=.009). 
Several of the TCC capacity indicators were significant predictors of TCC use. Those TCCs deemed to 
have more adequate facilities had higher TCC reporting than those with less adequate facilities in model 
 
25 The inclusion of these variables does introduce a slight autocorrelation problem as multiple precincts share the same TCC. 
3.3
2.7
3.9 3.5
0.0
1.0
2.0
3.0
4.0
5.0
Baseline Endline
Full Sample
Treatment Effect = 0.2pp
Control Treatment
3.3 3.2
4.6 4.8
0.0
1.0
2.0
3.0
4.0
5.0
6.0
Baseline Endline
Reduced Sample
Treatment Effect = 0.3pp
Control Treatment
29
1 (p=.032). Contrary to expectation, those TCCs that reported having one or more staff positions unfilled 
at the time of interview had statistically significantly higher TCC use, on average, than those fully staffed 
TCCs in model 4 (p=.062). This is likely a case of reverse causality, as high usage in some TCCs has 
generated an unfulfilled need for new staff. Also surprising is that those TCCs perceived to be meeting 
the needs of survivors had lower TCC use, which is perhaps indicative of those TCCs with higher 
caseloads finding it more difficult to meet the needs of survivors. 
Table 6: Regression results for impact of Treatment 1 on the logged average number of survivors 
using the TCC per month
(1) (2) (3) (4)
VARIABLES
Full sample, 
no controls
Reduced 
sample, 
no controls
Full 
sample
Reduced 
sample
DiD Treatment Effect (0-1) -0.12 -0.19 -0.12 -0.15
(0.29) (0.36) (0.27) (0.31)
Treatment Status (0-1) 0.32 0.52** 0.33* 0.45**
(0.21) (0.26) (0.19) (0.22)
Data collection period (0-1) -0.016 0.14 -0.021 0.14
(0.21) (0.25) (0.19) (0.21)
TCC understaffed (0-1) 0.041 0.48*
(0.17) (0.25)
TCC operates 24 hours (0-1) -0.14 0.13
(0.18) (0.23)
Facilities assessment (1-4) 0.20 0.44**
(0.15) (0.20)
Supplies assessment (1-4) 0.083 -0.16
(0.092) (0.13)
Space assessment (1-4) 0.10 0.15
(0.090) (0.11)
Meets survivor needs assessment (1-4) -0.39** -0.35
(0.18) (0.24)
Population size (log) 0.43*** 0.47***
(0.10) (0.12)
Sex crimes per 100k (log) 0.32* 0.17
(0.17) (0.20)
Murders per 100k 0.011** 0.013***
(0.0043) (0.0049)
Constant 0.98*** 0.97*** -5.68*** -6.02***
(0.15) (0.18) (1.40) (1.63)
Observations 142 90 142 90
Adjusted R-squared 0.005 0.032 0.182 0.282
Note: Standard errors in parentheses. Unit of analysis is a police precinct.
*** p<0.01, ** p<0.05, * p<0.1
30
3.4.4 Impact of service provider training on TCC utilization
Multiple regressions analysis using the difference-in-difference (DiD) method was also carried out to test 
the effect of service provider trainings (Treatment 2) following the same procedures and using the same 
four model specifications. As above, the DiD treatment effect variable in Table 7, our variable testing the 
treatment hypothesis, shows no impact of the service provider training on TCC use across all four models. 
Figure 12 shows that, at baseline, there were, on average, 4.7 survivor cases from control precincts and 
3.0 cases from Treatment 2 precincts under the full sample model specification, and this dropped to 3.5 
and 2.4 cases, respectively, at endline. As above, this decline is believed to be a drop in compliance in the 
use of supplemental intake forms and not necessarily indicative of a decline in TCC use. As this decline in 
compliance should affect the treatment and control precincts equally, we are still able to offer a test of 
the service provider training hypothesis. This DiD estimate indicates a small, positive gain of 0.6 cases per 
month in Treatment 2 precincts relative to control precincts from baseline to endline, though as previously 
noted, this difference is not statistically significant.
Figure 12: Difference-in-difference estimate of the effect of service provider training on TCC use
Precinct population size is a strong and statistically significant predictor in the model 3 regression (p=.003) 
as is the rate of sex crimes (p=.040), though these variables are insignificant in model 4. Three of the TCC 
variables statistically significantly predict TCC use in model 4: supplies assessment (p=.035), space 
assessment (p=.008), and if the TCC operates 24 hours a day (p=.070). 
4.7
3.5
3.0
2.4
0.0
1.0
2.0
3.0
4.0
5.0
Baseline Endline
Full Sample
Treatment Effect = 0.6pp
Control Treatment
3.9 4.0
3.1 2.9
0.0
1.0
2.0
3.0
4.0
5.0
Baseline Endline
Reduced Sample
Treatment effect = -0.3pp
Control Treatment
31
Table 7: Regression results for impact of Treatment 2 on the logged average number of survivors 
using the TCC per month
(1) (2) (3) (4)
VARIABLES
Full sample, 
no controls
Reduced 
sample, 
no controls
Full sample Reduced 
sample
DiD Treatment Effect -0.065 -0.28 -0.054 -0.29
(0.29) (0.36) (0.28) (0.33)
Treatment Status -0.19 -0.020 -0.18 -0.054
(0.20) (0.25) (0.20) (0.24)
Data collection period -0.035 0.29 -0.043 0.27
(0.21) (0.25) (0.20) (0.23)
TCC understaffed -0.11 0.28
(0.19) (0.28)
TCC operates 24 hours -0.071 0.44*
(0.19) (0.24)
Facilities assessment (1-4) 0.032 0.33
(0.14) (0.21)
Supplies assessment (1-4) 0.037 -0.29**
(0.10) (0.13)
Space assessment (1-4) 0.12 0.32***
(0.098) (0.12)
Meets survivor needs assessment (1-4) -0.16 -0.27
(0.18) (0.25)
Population size (log) 0.34*** 0.20
(0.11) (0.14)
Sex crimes per 100k (log) 0.37** 0.30
(0.18) (0.20)
Murders per 100k 0.0020 0.0034
(0.0038) (0.0043)
Constant 1.21*** 1.05*** -4.34*** -3.23*
(0.15) (0.18) (1.51) (1.85)
Observations 148 95 148 95
Adjusted R-squared -0.002 -0.009 0.078 0.146
Note: Standard errors in parentheses. The unit of analysis is the police precinct.
*** p<0.01, ** p<0.05, * p<0.1
In summary, supplemental intake data collected from the TCC show that TCC use is low, with a median
of 12.7 survivor cases per month per TCC across the evaluation period. Both the full and reduced samples 
of control and treatment precincts were balanced at baseline on the dependent variable as well as relevant 
control variables. Regression analyses using difference-in-difference show no impact of the community 
dialogues or service provider training on TCC use. Statistically significant explanatory variables for TCC 
use include population size, rate of sexual offences, and rate of murders. There is also some indication 
that TCC capacity influences utilization; however, the effect of various measures of TCC capacity, such as 
staffing, after-hours operation, facilities, and supplies, is not consistent across the models and should be 
interpreted carefully. 
32
3.5 IMPACT OF COMMUNITY DIALOGUES ON COMMUNITY KNOWLEDGE 
AND ATTITUDES
Women’s survey data was collected at households within the selected communities near each TCC to 
measure women’s knowledge of TCCs and SGBV, and attitudes toward SGBV. The women’s survey was 
administered to a cross-section of 1,500 women across Treatment 1 (community dialogue) and control 
police precincts in all nine provinces in South Africa before and after community dialogues, which were 
implemented by Sonke and Soul City. The evaluation tests whether communities that received outreach 
efforts realized a positive change in knowledge and attitudes vis a vis the control communities. Table 8
shows the distribution of women’s survey respondents by province. 
Table 8: Number of women’s survey respondents by province 
Province Frequency Percentage Cumulative
Eastern Cape 480 16 30
Free State 240 8 38
Gauteng 420 14 14
Kwa-Zulu Natal 420 14 52
Limpopo 300 10 62
Mpumalanga 300 10 72
North West 270 9 81
Northern Cape 270 9 90
Western Cape 300 10 100
Total 3,000 100% 100%
Note: Including both baseline and endline. The sample is not intended 
to make national or provincial generalizations.
3.5.1 Balance statistics
In randomized samples of individuals, people are randomly selected into control and treatment groups 
such that—provided a large enough sample size—any differences observed between the two groups would 
be a product of random chance. However, because random assignment was at the precinct level and since 
there were only a limited number of precincts, there was a risk that sampled Treatment 1 and control 
groups would be systematically different from one another on key characteristics. Fortunately, the balance 
statistics of the final sample confirm that the selection and random assignment process was successful in 
creating comparable treatment and control groups. 
Independent sample t-tests and chi-square analyses were conducted to test for statistical equivalence of 
the treatment and control groups. The results are presented in Table 9. While several small but statistically 
significant differences were observed between the treatment and control groups in the baseline sample, 
the treatment and control groups were statistically equivalent on all demographic variables in the endline 
sample. 
 
33
Table 9: Balance statistics between endline Treatment 1 and control samples (n=1,500) 
Variable Treatment 
(%)
Control 
(%) p-value
Age 33 33 0.815
Black 86 85 0.509
Coloured 11 11 0.805
Zulu 19 19 0.793
Xhosa 19 19 0.895
Tswana 16 15 0.667
Afrikaans 12 12 0.937
Other language 33 35 0.355
Less than elementary education 1 1 0.615
Elementary 23 22 0.757
Secondary 67 67 0.826
Higher than secondary 9 10 0.541
Monthly income less than or equal to 1,000 R 8 9 0.712
Monthly income between 1,000-5,000 61 59 0.399
Monthly income higher than 5,000 31 32 0.541
The sample of women surveyed for this IE should not be considered representative of all women in South 
Africa, but rather representative of women in evaluation precincts. Furthermore, the survey eligibility 
criteria stipulated that respondents were to be between the ages of 18 and 49 and reside in areas not
identified as high-income areas (in which the majority of residents have private insurance), and as such, 
are not the targeted beneficiaries of the TCCs. The demographic characteristics of the endline sample are 
shown in Figure 13. On average, women in the endline sample were 32.5 years old, identified themselves 
as black, had attained a secondary education, and reported a household income of 1,001 to 5,000 Rand 
($66 to $329 USD) per month.
34
Figure 13: Endline respondent demographics (n=1,500)
3.5.2 Knowledge of TCCs and other services in the community to victims of sexual assault
A formative research study conducted by Soul City Institute in 2013 on knowledge, attitudes, and 
behaviors relating to sexual assault reporting and the use of TCCs found that the vast majority of South 
Africans are not aware of TCCs. The baseline women’s survey data mirrors the results of this study. At 
baseline, only 18 percent of respondents reported knowledge of the TCC.
At the beginning of each baseline and endline survey, respondents were asked to identify services to 
victims of sexual assault in their communities, and their providers. At baseline, 21 percent of respondents 
said there are services to victims of sexual assault in their communities, compared to 20 percent at endline. 
Those respondents who reported knowledge of such services were then asked to identify the service 
providers without being prompted. Most respondents were not able to identify a provider, and few 
mentioned the TCC, although more respondents mentioned the TCC at endline compared with baseline. 
As shown in Figure 14, there were no meaningful differences in service provider identification between 
the treatment and control areas. 
86%
2%
11%
1%
Respondent race
Black
White
Coloured
Indian
2%
23%
61%
15%
Level of formal education
No formal education
or only pre-school
Elementary
Secondary
Tertiary or
Graduate
16%
58%
20%
6%
Household income last month
Less than R1,000
R1,001-5,000
R5,001-10,000
More than
R10,001
11%
3% 4%
7%
10%
5%
3%
15%
3%
20%20%
0%
5%
10%
15%
20%
25%
Primary language at home
35
Figure 14: Identification of service providers to victims of sexual assault (n=302 at baseline and 279 
at endline)
The respondents who reported knowledge of services for SGBV survivors were also asked to identify the 
types of support provided in the community to victims of sexual assault. At endline, the share of 
respondents in the treatment group who knew of medical services was 14 percent, emotional/
psychological services was 12 percent, and legal services was eight percent. The treatment group was no 
more knowledgeable than the control group with regards to services provided in the community.
Figure 15: Type of support services provided in the community (n=301 at baseline and 277 at 
endline)
When women were asked where they would go first to report a sexual assault, less than one percent said 
they would first report a sexual assault to a TCC and more than 70 percent at endline cited the police as 
the first point of reporting. Qualitative data reveal that most victims brought to the TCCs are brought by 
the police. As such, reporting to the police should result in the survivor ultimately arriving at the TCC for 
services; however, as we know, many survivors do not report SGBV to the police. 
 
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
Baseline Endline Control Endline Treatment
6%
11%
8%
15% 15%
7%
14%
12%
8%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Medical Emotional Legal
Baseline Endline Control Endline Treatment
36
Figure 16: Respondents’ first response if victim of sexual assault (n=1,499 at baseline and 277 at 
endline)26
As shown in Table 10, knowledge of TCCs is relatively low, with between 16 percent and 19 percent of 
respondents reporting awareness of the TCCs across condition groups and time. Simple bivariate 
regression analysis using a difference-in-difference model was conducted to measure changes in knowledge 
of TCCs and resources available to survivors of SGBV in treatment and control communities from baseline 
to endline. At endline, levels of knowledge of TCCs in sampled communities is statistically similar to 
baseline (p=0.606), indicating no positive general time trend in awareness of TCCs or services available 
to survivors of SGBV across these communities over the evaluation period. Moreover, knowledge levels 
between treatment and control communities are also statistically equivalent at endline within both models: 
using a first difference (p=0.840) and difference-in-difference (p=0.903) regression model, indicating the 
absence of a positive treatment effect of the community dialogues on knowledge of TCCs or services to 
survivors of SGBV. Summarized results are presented in Table 10. Treatment effects from regression 
results are presented as odds ratios.27 All odds ratios are at or near one, indicating no treatment effect.
 
26 At endline, this question was only administered to those who answered “yes” to “are there services available in your 
community for victims of sexual assault?”, thus, the large discrepancy in sample size and in results. 27 Odds ratios are another way to interpret the coefficients of a logistic regression; odds are defined as the ratio of 
the probability of success and the probability of failure. For example, an odds ratio greater than 1 indicates that the
odds are more likely that the treatment had a positive effect on a given outcome, whereas an odds ratio less than 
1indicates that the treatment likely had a negative effect on a given outcome.
0%
10%
20%
30%
40%
50%
60%
70%
80%
Baseline Endline Control Endline Treatment
37
Table 10: Respondent has heard of the TCC (n=1,500 at baseline and endline) 
Treatment Control Diff￾in-diff
Treatment 
effect* p-value
 Baseline Endline Baseline Endline
Has heard of the 
Thuthuzela Care Center 
(TCC)
17% 16% 19% 17% 1pp 1.03 0.903
Has heard of a place like 
the TCC in the community 2% 3% 5% 4% 2pp 1.16 0.542
Knows of services available 
in her community for 
survivors of GBV
20% 20% 22% 21% 1pp 0.96 0.882
Knows of care centers run 
by NGOs 20% 17% 21% 17% 1pp 1.09 0.874
 * Odds Ratio
The sample was not designed to be representative at the province level; but it is possible that the effects 
of the intervention vary geographically. Province-disaggregated regression analyses suggest variation in 
treatment effects across provinces; however, the observed differences are generally statistically 
insignificant. Women surveyed in treatment communities in North West, for example, were 2.08 times 
as likely to have heard of the TCC than women surveyed in control communities at endline; however, 
this relationship is statistically insignificant (see Table 11). Conversely, women in treatment communities 
in Western Cape were significantly less likely (OR=.33) to have heard of the TCC than women in control
communities at endline (p=.065).28
Table 11: Knowledge of TCCs by province (n=1,500 at baseline and endline) 
Province
Treatment Control Diff-in￾diff
Treatment 
effect+ p-value
n Baseline Endline Baseline Endline
Gauteng 210 2% 3% 2% 5% -2pp 0.56 0.675
Eastern Cape 240 34% 23% 41% 21% 8pp 1.35 0.514
Free State 120 37% 13% 33% 12% -2pp 1.09 0.890
Kwa-Zulu Natal 210 10% 24% 6% 31% -11pp 0.42 0.336
Limpopo 150 15% 32% 9% 19% 8pp 0.83 0.860
Mpumalanga 150 8% 8% 7% 5% 1pp 1.16 0.865
North West 135 20% 19% 25% 17% 7pp 2.08 0.269
Northern Cape 135 20% 12% 24% 15% 1pp 1.54 0.476
Western Cape 150 13% 8% 23% 24% -7pp 0.33 0.065*
 + Odds Ratio
Figure 17 shows logistic regression results for the effect of the community dialogues on knowledge of 
TCCs while controlling for province, demographic characteristics, measures of sexual assault prevalence, 
and other explanatory outcome variables. Each variable in the figure has a dot that indicates the estimated 
effect of the independent variable on the odds (technically the logged odds) of knowing of a TCC. The 
blue line surrounding the dot indicates the confidence interval, which considers random error in the data. 
 
28 One potential explanation for the considerable variation across provinces is that Thuthuzela is a Xhosa word, which would be 
most familiar in Xhosa speaking areas such as the Eastern Cape.
38
If the blue line overlaps the red line for a particular variable, this means that we cannot reasonably rule 
out the chance that there is no relationship between the variable and knowledge of the TCC. When the 
blue line does not overlap the red line, then we are reasonably confident that a relationship exists. The x￾axis of the figure is an odds ratio. When the odds ratio equals one then a change in the value of 
independent variable is not expected to produce a change in the odds of knowing a TCC. With the value 
rises to a two it means that a change in the value of the independent variables produces a doubling of the 
odds of knowing a TCC. An odds ratio of less than one implies a negative relationship. Figure 17 confirms 
the finding above, that treatment is not a significant predictor of familiarity with the TCCs. 
While the sample is not intended to be representative at the provincial level, province is the strongest
and most significant predictor of familiarity with the TCCs; however, we have opted not to include 
province in the figure because the effect sizes are so large. As shown in Table 11, knowledge of the TCCs 
was lowest among respondents in Gauteng. Sampled women in Eastern Cape were 17.1 times more likely 
to know of the TCC than women in Gauteng. Those in Limpopo were 7.7 times more likely, those in Free 
State were 7.2 times as likely, Northern Cape 6.5 times more likely, Western Cape 5.9 times more likely, 
North West 5.3 times more likely, Kwa-Zulu Natal 4.5 times more likely, and Mpumalanga, 2.6 times 
more likely than women in Gauteng to know of the TCC. 
Sexual assault prevalence and perceptions of sexual assault as a problem also significantly predict 
knowledge of the TCC. Women residing in precincts with higher rates of sexual assault crimes are 1.3 
times more likely to know of the TCC (p=.017). Similarly, women who reported knowing someone who 
has been raped or sexually assaulted were 4.8 times more likely to know of the TCC (p<.001), and women 
who perceived SGBV as a problem in their communities were 1.1 times more likely to know of the TCC 
(p=.089). 
Women with low tolerance for violence toward women were also slightly more likely to know of the 
TCC (p=.023). Several demographic variables were statistically significant predictors of knowledge of the 
TCC: age (p<0.05), education (p<0.001), income (p<0.05), and primary language (p<0.01). Women who 
are older, have a secondary level of education or above, and have higher household income are more 
likely to have heard of the TCC. At baseline, knowledge of TCCs was also significantly associated with 
age (p<.01), income (p<.05), and primary language (p<0.001), again with older and higher-income 
respondents being more likely to know of the TCCs. Education was statistically significant at baseline only 
at the 10 percent level, with a p-value of 0.095. Full regression tables are included in Annex VI. 
39
Figure 17: Logistic regression results for effect of Treatment 1 on knowledge of TCCs (n=2,950)
The 294 (16.5 percent) endline respondents who did report knowledge of TCCs or knowledge of a similar 
place in their communities were asked a series of additional questions about TCCs and TCC services. 
The majority of these respondents reported knowing of the services the TCC offers (88 percent) and the 
location of the nearest TCC (93 percent). Two follow-up questions were asked to assess possible 
response bias on these two items. The first question asked respondents to name the location of the 
nearest TCC: 258 out of 260 respondents who reported knowing the location of the nearest TCC were 
able to provide a specific location when asked, and 198 (76 percent) of these responses were correct.
Figure 18 shows that there is no significant treatment effect across knowledge of services nor knowledge 
of location in Treatment 1 communities as compared to control communities. 
DiD Treatment effect
Treatment status
Data collection period
Age
Zulu
Xhosa
Tswana
Afrikaans
Coloured
Lower than secondary education
Tertiary education or higher
Income<R1,000 per month
Income>R5,001 per month
Urban precinct
Knows a SGBV survivor
Perception of SGBV as a problem in the community
Sex crimes per 1,000
Woman's right to refuse sex PCA
Tolerance of violence towards women PCA
Rape victim-blaming PCA
0 2 4 6
Odds ratio
40
Figure 18: Share of respondents who said they know about TCC services and location (n=312 at 
baseline and 294 at endline) 
The second follow-up question enquired about types of services offered by the TCC. Nearly all 
respondents who indicated knowledge of the services offered by TCCs responded correctly that TCCs
offer medical and psychological services, and the vast majority reported knowledge of provision of legal 
services (Table 12). Few respondents reported knowledge of transportation services, both at baseline and 
at endline, though this did increase statistically significantly at endline (p=.069). While transportation to 
the TCC is provided, transportation home from the TCC or back to the TCC for follow-up appointments 
is not. Moreover, transportation to the TCC is only offered from the police station, so respondents may 
perceive transportation as being offered by the police rather than the TCC. 
Table 12: Knowledge of TCC services (n=244 at baseline and 223 at endline)
 Treatment Yes (%) Control Yes (%) 
Knowledge of TCC 
services
Correct 
response
Baseline Endline Baseline Endline Diff-in￾diff
Transportation Yes 7% 41% 13% 30% 17pp* 
Medical assistance Yes 98% 99% 100% 97% 3pp
Psychological and counseling 
service Yes 99% 99% 100% 98% 2pp
Legal assistance Yes 82% 88% 78% 91% -8pp
Important misperceptions about the TCCs remained at endline; 87 percent of treatment respondents 
who claimed familiarity with TCCs or a similar place in their community incorrectly answered that clients 
visiting the TCC must report the name of her/his attacker, and 93 percent incorrectly believed that the 
TCC requires her/him to take legal action. These misperceptions are likely to dissuade victims from 
utilizing TCC services. Unfortunately, the intervention does not appear to have reduced these 
misperceptions (See Table 13). At endline, there had been minimal reductions in these misperceptions in 
the treatment communities, and respondents in treatment areas were actually less likely to know that 
people under the age of 18 are eligible to receive help from the TCC.
79%
69%
76%
90%
75%
86%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Know of services Know of location
Baseline Endline Control Endline Treatment
41
Table 13: Specific TCC knowledge (n=292 at baseline and 282 at endline)
Treatment Agree (%) Control Agree (%)
 Baseline Endline Baseline Endline Diff-in￾diff
A man can receive help from the TCC 84% 88% 82% 89% -3pp
A child can receive help from the TCC 90% 57% 86% 52% 1pp
Clients do not have to pay for TCC services 96% 92% 98% 89% 5pp
Clients are not required to report the name of 
the attacker 23% 13% 20% 19% -9pp
It is possible to file a police report at the TCC 77% 91% 80% 90% 4pp
The TCC does not require you to take legal 
action 12% 7% 14% 9% 0pp
To measure community awareness of TCCs, respondents in the subsample (those who have heard of the 
TCCs) were asked their perception of community knowledge of TCCs. As shown in Figure 19, few believe 
that most or all women are aware of these services. At endline, 87 percent of treatment respondents, and 
88 percent of control respondents, believe that none or a few women know about the TCC.
Figure 19: Respondent perception of community awareness of TCCs (n=283 at baseline and 194 at 
endline)
It was hypothesized that, through community dialogues, knowledge of the TCCs would spread throughout 
the community. As presented in findings throughout this section, women in treatment precincts generally 
did not have higher levels of knowledge about the TCCs after the community dialogues than women in 
control precincts. Thus, we are not able to reject the null hypothesis that the dialogues had no effect on 
women’s knowledge in communities. 
One possible explanation for these null findings is that the community dialogues did not reach enough 
women in the treatment precincts. Two survey questions were asked to estimate the programming’s 
reach, and the results provided supporting evidence to this hypothesis. When asked whether in the past
three months they had heard any announcements about sexual assault awareness or about resources 
18 19
30
55
69
57
23
11 6
4 1 6
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Control Treatment
Baseline Endline
In thinking of the adult women in your neighbourhood, 
how many of them do you think know about the TCC?
Most/all
Some
A few
None
42
available to victims of sexual assault, respondents were no more likely to report having heard these 
messages at endline than baseline (see Figure 20), and those in the treatment group were no more likely 
to report awareness than those in the control group. Messages conveyed in the dialogues appear to not 
have spread widely enough in communities to have a meaningful impact on community knowledge. This 
finding has important policy implications. Even if Treatment 1 was very effective in increasing knowledge 
for dialogue attendees, we do not find evidence that such an approach is adequate in reaching the larger 
population and increasing community knowledge about TCCs. 
Figure 20: Share of respondents who heard ads/announcements about sexual assault in the last
three months (n=1,495 at baseline and 1,498 at endline)
3.5.3 Perceptions of sexual assault and sexual assault reporting
To compare how women perceive sexual assault relative to other crime problems, respondents were 
asked to rate several common crime problems on a four-point scale—not a problem, a minor problem, a 
problem, or a major problem. Both at baseline and at endline, respondents were more inclined to report 
house-breaking or mugging as problems than sexual assault or domestic violence, even though sexual 
assault might be more prevalent (see Figure 21). According to SAPS crime statistics there were 53,439 
instances of reported common robbery in South Africa in 2013 compared with 66,197 instances of 
reported sexual assault.29 While both are likely to be underreported, the baseline findings suggested that 
it was necessary to continue to raise awareness about the extent and gravity of the problem of sexual 
assault. As expected, respondents in precincts with higher rates of reported sexual assault were 
statistically significantly more likely (p<.01) to indicate sexual assault as “a problem” or a “major problem” 
than “not a problem” or “a minor problem.” 
 
29 South African Police Service (2014) Crime Situation in South Africa. 
http://www.saps.gov.za/resource_centre/publications/statistics/crimestats/2014/crime_stats.php 
42%
24% 29%
22% 26%
20%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Awareness of sexual assault Awareness of resources available to
victims of sexual assault
In the last 3 months, have you heard announcements 
promoting...
Baseline Endline Control Endline Treatment
43
Figure 21: Respondent perception of crime problems at endline (n=1,500)
At endline, despite over half of respondents ranking sexual assault as “not a problem” or “a minor 
problem,” 17 percent of respondents report personally knowing women or girls who have been raped or 
sexually assaulted in the last year, and 22 percent report personally knowing a woman or girl who has 
been raped or sexually assaulted in her lifetime (see Figure 22). On average, this sub-sample of respondents 
reported knowing 2 women or girls who had been assaulted in the last year, 2.7 women or girls who had 
been assaulted in their lifetime, and 0.7 men and 1.4 boys who had been assaulted in the last year. Noting 
the decline in reporting over time we opted to examine this data further using logistic regression. The 
results (not presented here in full detail) show that across both treatment and control precincts, women 
were less likely to report knowing a victim of SGBV at endline, which is statistically significant across all 
three categories, indicating a possible decrease in incidents over the evaluation period. However, the 
difference-in-difference model finds no statistically significant effects in Treatment 1 precincts when 
compared with control precincts, indicating that this decrease cannot be attributed to the community 
dialogues. 
44
Figure 22: Share of respondents who know SGBV victims
Respondents in the subsample familiar with TCCs or similar places in their communities were asked how 
many women, men, girls, and boys they know personally who have visited a TCC. Results are presented 
in Table 14 and show that these respondents reported the highest rate of TCC use among girls under the 
age of 18. Endline respondents were less likely to know SGBV victims who have visited a TCC, irrespective 
of the victims’ age, although this trend is observed both in treatment and control areas. 
Table 14: Number of SGBV survivors the respondent knows who have visited a TCC (n=283 at 
baseline and 194 at endline)
How many... do you personally know 
that have visited a TCC? 
Treatment (mean) Control (mean) 
Baseline Endline Baseline Endline Diff-in￾diff
Women 18 and older 1.32 0.14 1.19 0.18 -0.16
Men 18 and older 0.15 0.01 0.15 0.00 0.01
Girls 18 and under 1.65 0.16 1.39 0.13 -0.23
Boys 18 and under 0.80 0.01 0.56 0.02 -0.25
To increase utilization of the TCCs, it is important to understand the barriers to use. Respondents were 
asked to categorize a series of possible barriers to visiting a TCC as either: not at all a barrier, a minor 
barrier, a barrier, or a major barrier. At baseline, nearly three quarters of respondents reported that (lack 
of) awareness of the TCCs was a major barrier to visiting TCCs, which was the most common barrier 
reported. These results made clear the need for greater outreach to the communities about the TCCs 
and their services. At endline, a lack of awareness was still cited as the greatest barrier to visiting TCCs, 
with three quarters of respondents reporting it as such (see Figure 23). The second most prominent 
barrier was fear of punishment by the perpetrator, followed by the offer of money not to report the 
sexual assault. 
25%
37%
3%
19%
23%
9%
15%
21%
8%
0%
5%
10%
15%
20%
25%
30%
35%
40%
Women/girls (last year) Women/girls (lifetime) Men/boys (last year)
Baseline Endline Control Endline Treatment
45
Figure 23: Respondent perceptions of barriers to visiting the TCCs (n=1,500)
Using the same structure as the previous question, respondents were asked to categorize a series of 
possible barriers to reporting a sexual assault case to the police. At baseline, respondents reported fear 
of punishment by the perpetrator and a fear that the perpetrator would find out as the most significant 
barriers. At endline, fear of punishment by the perpetrator and money offered not to report were 
identified as the most serious barriers. Comparing the two figures reveals both the strengths and 
limitations of the TCC vis-à-vis the police. Fear that that they will not be supported, not be believed, or 
will not have anyone to trust is higher when asked about the police than when asked about the TCCs. 
Nonetheless, respondents reported similar level of fear that the community will find out about the assault 
or that the perpetrator will find out the assault was reported. 
75
63
59
55
40
39
35
32
28
7
7
7
17
13
12
17
32
27
32
13
19
26
24
28
3
9
11
12
12
15
15
14
23
34
34
32
4
15
18
17
15
19
17
41
30
33
35
34
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Not aware of centers
Fear of punishment by perpetrator
Money offered not to report
Fear perpetrator would find out
Feeling ashamed or embarassed
Fear of victim-blaming
Fear community would find out
TCCs not open at convenient times
Transportation challenges
Feeling there is no one to trust
Fear of others at TCC not believing
Fear of not receiving support
Major Barrier Barrier Minor barrier Not a barrier
46
Figure 24: Respondent perceptions of barriers to reporting to the police at endline (n=1,499)
3.5.4 Attitudes toward gender
The women’s survey also included several questions drawn from other international surveys designed to 
assess attitudes toward gender and gender roles. Specifically, items were drawn from the 2008 
International Men and Gender Equality Women’s Survey and the 2003 World Health Organization 
(WHO) Multi-country study of Women’s Health and Life Events. Generally speaking, the results showed
progressive gender attitudes among respondents at baseline. Very few respondents indicated scenarios 
when violence toward woman would be warranted or tolerated. Moreover, responses indicated low levels 
of victim-blaming in rape cases. Changes in observed attitudes between baseline and endline are show in 
Table 15. None of the observed difference in difference comparisons were statistically significant. 
63
62
55
54
40
39
36
34
33
32
17
1
14
12
21
12
29
35
33
27
27
29
12
3
13
12
15
11
18
18
19
14
14
15
17
9
9
15
9
23
13
8
11
26
27
23
55
86
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Fear of punishment by perpetrator
Money offered not to report
Fear perpetrator would find out
Lack of information
Fear of victim-blaming
Feeling ashamed or embarassed
Fear community would find out
Fear of not receiving support
Fear of police not believing
Feeling the police cannot be trusted
Transportation challenges
Station not open at convenient times
Major Barrier Barrier Minor barrier Not a barrier
47
Table 15: Percentage of respondents indicating agreement with gender statements (n=1,500 at 
baseline and endline)
 Treatment Control Diff-in￾diff
Desired & 
Observed 
Gender Attitudes Baseline Endline Baseline Endline Change
Rights for women mean that men lose 
out.
4.9 5.1 6.2 3.5 2.9 ↓↑
There are times when a woman 
deserves to be beaten. 4 2.1 3.9 2.1 -0.1 ↓↓
A man should have the final word 
about decisions in his home. 18.6 10 16.5 10.3 -2.4 ↓↓
A woman should tolerate violence in 
order to keep her family together. 9.4 5.7 6.1 4.9 -2.5 ↓↓
A man and a woman should decide 
together what type of contraceptive to 
use. 
48 33 45.1 32.7 -2.6 ↑↓
If a man sexually assaults his wife, 
others outside of the family should 
intervene.
70.2 64.3 72.1 61.6 4.6 ↑↑
 
A man can hit a woman if… 
She does not complete her housework 
to his satisfaction. 0.9 0.1 0.7 0.3 -0.4 ↓↓
She disobeys him. 4.6 2.8 2.3 3.5 -3 ↓↓
She refuses to have sexual relations 
with him. 1.9 0.9 1.6 1.9 -1.3 ↓↓
He suspects that she is unfaithful. 1.9 0.7 0.8 1.1 -1.5 ↓↓
He finds out that she has been 
unfaithful. 9 4.8 5.1 5.9 -5 ↓↓
 
A woman can refuse sex to her husband if… 
She doesn't want to. 55.2 81.6 60.3 85.5 1.2 ↑↑
He is drunk. 65.3 87.2 67.4 90.1 -0.8 ↑↓
She is sick. 75.4 96.1 76.9 96.1 1.5 ↑↑
He mistreats her. 74.3 96.8 76.2 96.3 2.4 ↑↑
 
When a woman is raped, she usually 
did something careless to put herself in 
that situation.
7.6 4.5 4.1 3.2 -2.2 ↓↓
In some rape cases, women actually 
want it to happen. 6 5.7 4.7 4.8 -0.4 ↓↓
If a woman doesn't physically fight back, 
you can't really say it was rape. 23.2 2.4 23.2 2.1 0.3 ↓↑
In any rape case, one would have to 
question whether the victim sleeps 
around a lot or has a bad reputation. 
28.3 32.6 28.3 28.8 3.8 ↓↑
48
Women were also asked to respond to a series of questions following two hypothetical scenarios of 
sexual assault. The purpose of these questions is to assess the respondent’s attitudes toward rape in two 
distinct scenarios: 
Scenario 1: A woman has to work late each night. The bus she takes home lets her off .5km from her 
home. One night, when walking home, she is assaulted by a man. She is unable to fight him off and he 
rapes her. 
Scenario 2: An attractive 20-year-old single woman wearing a mini-skirt goes out on a Friday night with 
friends. She stays for a few hours and has a few drinks. On her way home, she is assaulted by a man. She 
is unable to fight him off and he rapes her. 
At baseline, respondents were more likely to agree that the woman was partially to blame for being 
sexually assaulted in Scenario 2 than in Scenario 1. Moreover, respondents reported that women in their 
neighborhood would be less likely to file a police report and that the police would be less helpful for 
Scenario 2. As shown in Figure 25, respondents in the treatment group were no more likely to respond 
favorably to either of these scenarios. 
Figure 25: Prevalence of victim-blaming attitudes among respondents at endline, by scenario
(n=1,500)
A SGBV attitudes index was constructed by combining responses from 21 survey items measuring 
attitudes toward gender roles, and the sexual assault scenarios, using Principal Component Analysis (PCA). 
A higher PCA score indicates progressive gender attitudes and lower tolerance of SGBV. We then 
conducted a multiple regression analysis to test whether the treatment produced higher PCA scores.
Additional variables were added to the analysis to increase precision and better understand variation in 
SGBV attitudes. The DiD treatment effect variable provides a test of the effect of treatment. The estimate 
shows an improvement in attitudinal scores in the treatment group relative to the control group from 
49
baseline to endline, although this relationship is not statistically significant (p=.235). As such, we cannot 
conclude that the intervention had an effect on attitudes in intervention communities. 
Variables that are statistically significant and positively associated with more normatively desirable 
attitudes scores are: age (p=.002), Zulu (p=.002), higher income (p<.001), and knowing a woman or girl 
who has been raped or sexually assaulted (p<.001). 
Figure 26. Regression results for impact of community dialogues on attitudes toward SGBV
(n=2,931)
In summary, while the survey reveals that most respondents have normatively desirable attitudes toward 
gender roles and the sexual assault scenarios, some respondents still that engage in victim blaming and 
there is still evidence to suggest that social stigmas represent a barrier to TCC usage. Nonetheless, as 
with knowledge of TCCs above, there is no evidence that the intervention had an impact on community 
attitudes toward gender roles and sexual assault. As concluded above, even if Treatment 1 was very 
effective among those that attended the dialogues, such an approach is inadequate to reach the larger 
population and change community attitudes. 
DiD Treatment effect
Treatment status
Data collection period
Age
Zulu
Xhosa
Tswana
Afrikaans
Lower than secondary education
Tertiary education or higher
Income<R1,000 per month
Income>R5,001 per month
Urban precinct
Sex crimes per 1,000
Perception of SGBV as a problem in the community
Knows a SGBV survivor
-1 -.5 0 .5 1 1.5
50
3.6 IMPACT OF TRAINING ON SERVICE PROVIDERS’ KNOWLEDGE AND 
ATTITUDES
Pre- and post-training surveys were administered to service providers attending the FPD-led Integrated 
Management of Sexual and Gender-Based Violence (Integrated Management) training (Treatment 2). All 
training participants were requested to consent to and complete a brief, self-administered survey about 
their knowledge of TCCs and SGBV, perceptions, and recent practices with survivors of SGBV before the 
start of the first day of training. The evaluation team followed up with each service provider approximately 
three months later to test knowledge gains and changes in perceptions and practices with survivors of 
SGBV. 
Table 16 shows the number of professionals attending the training in each province, the number of those 
professionals agreeing to participate in the evaluation at baseline, and the number reached for the follow￾up survey at endline.30 The data collection firm, IRI, attempted to reach each service provider surveyed at 
baseline for the follow-up survey. In total, eight attempts were made to reach a respondent for the endline 
survey. The following data analysis is conducted on the sub-sample of attendees for whom both a baseline 
and an endline survey were collected, yielding a sample size of 1,347 respondents, which represents 71 
percent of the trained professionals.
Table 16: Number of professionals trained, surveyed at baseline and at endline
Province # Trained # Surveyed 
at baseline
% Trainees 
Surveyed at 
baseline
# Surveyed 
at endline
% Trainees 
Surveyed at 
endline
Eastern Cape 377 356 94% 228 60%
Free State 155 148 95% 112 72%
Gauteng 222 212 95% 145 65%
Kwa-Zulu 
Natal* 276 245 89% 251 91%
Limpopo 253 233 92% 155 61%
Mpumalanga 138 138 100% 120 87%
North West 226 214 95% 165 73%
Northern Cape 118 104 88% 78 66%
Western Cape 143 139 97% 93 65%
Total 1,908 1,789 94% 1,347 71%
* Excludes participants from Margate precinct in KwaZulu-Natal province. This precinct was dropped from the 
evaluation due to delays in implementation. 
As envisioned in the training curriculum, trainees represented a variety of professions, with the majority 
being NGO workers, social workers, police officers and health workers. Figure 27 shows the share of 
respondents by professions for those who were interviewed at baseline and endline (n=1,308). Of those 
attending the training, 36 percent reported having previously attended a training on sexual assault (out of 
n=1,321). 
 
30 Ninety-four percent of trainees consented to participate in the baseline survey; however, respondents often skipped questions 
on the self-administered baseline survey. As such, sample sizes vary somewhat from question to question, although for the 
majority of questions, data is not missing for more than 10% of respondents. 
51
Figure 27: Respondent professions (n=1,308)
Three quarters of respondents in the sample who reported their sex (n=1,326) were female. Respondent 
ages are presented in Figure 28. The majority of respondents were under 40 years old.
Figure 28: Respondent age (n=1,330)
3.6.1 Knowledge of TCCs
Respondents were asked a series of questions to test their knowledge of TCCs. Respondents were fairly 
knowledgeable about the TCCs at baseline, with 80 percent knowing of services for SGBV survivors. This 
is perhaps reflective of the fact that participants self-selected to participate in the training, ostensibly 
because of an interest in the issue. At endline, service providers were statistically significantly more likely 
to report knowing what services were available for survivors of SGBV in their communities, the services 
provided by the TCCs, and the location of TCCs (see Table 17).
0%
0%
1%
1%
2%
4%
7%
7%
16%
17%
23%
23%
0% 5% 10% 15% 20% 25%
Church
Traditional leader
Court/legal
Government leader
TCC Site Coordinator
TCC Victim Assistance Officer
Educator
Other
Health worker
Police officer
Social worker
NGO worker
1%
27%
37%
24%
10%
1%
0% 5% 10% 15% 20% 25% 30% 35% 40%
<18
18-30
31-40
41-50
51-60
>60
52
Table 17: Knowledge of TCCs (n=1,347)
Respondents reporting… Baseline (%) Endline (%) pp 
difference
significant 
at 95%
Services are available in my 
community for survivors of SGBV 80 89 9 * 
I know what services the TCCs 
offer 81 99 18 * 
I know the location of the TCC 78 98 20 * 
Although most reported knowledge of TCC services, prior to the training, respondents did not have a 
good understanding of what services are offered at TCCs at baseline (Table 18). Before the training, just 
over half of respondents correctly identified that TCCs provide medical assistance, while 77 percent
correctly noted that TCCs provide psychological and counseling services. Notably, only 38 percent
believed that TCCs provide legal assistance to survivors, despite litigation being an important facet of TCC 
services. Finally, as was the case with women’s survey respondents, few professionals reported knowing 
about TCCs’ provision of transportation services, even though this service is provided for the TCCs via 
the police. 
After the training, the percentage of respondents who knew that these services are provided increased 
dramatically, with nearly all reporting that TCCs offer medical assistance, counseling, and legal assistance. 
It should be noted however, that almost half still did not answer correctly when asked if the TCCs provide 
transportation services. Improvements on each of these items is statistically significant at the 95 percent
confidence level. 
Table 18: Knowledge of TCC services (n=1,347) 
Knowledge of TCC services Correct response 
Transportation Yes
Medical assistance Yes
Psychological and counseling Yes
Legal assistance Yes
Table 19 shows responses to additional questions about TCCs before and after the training. Knowledge 
of these items also increased significantly from baseline to endline. After training, 99 percent of trainees 
were aware that men and those under the age of 18 can receive help from the TCC, and 95 percent knew 
that TCC services are free. Although respondents became better informed regarding these questions, 34
percent still believe that TCC clients are required to report the name of the attacker and 46 percent
believe that TCC clients are required to take legal action. 
53
Table 19: Knowledge of more specific TCC services (n=1,280 at baseline and 1,346 at endline) 
Survey question
% 
responding 
correctly 
at baseline
% 
responding 
correctly 
at endline
PP 
difference
significant 
at 5%
A man can receive help from the TCC 88 99 11 * 
A child can receive help from the TCC 89 99 10 * 
Clients do not have to pay for TCC services 72 95 23 * 
Clients are not required to report the name of the 
attacker 28 66 38 * 
It is possible to file a police report at the TCC 49 82 33 * 
The TCC does not require you to take legal action 27 54 27 * 
Respondents were also asked to report how widely known TCCs are at their workplace. As shown in
Figure 29, responses to this item varied substantially. There was high variability in TCC awareness between 
workplaces both across professions and within the same profession. After the training, respondents 
reported statistically significantly higher levels of TCC awareness in their workplaces, providing some 
evidence to suggest that the training might have a spillover effect on other colleagues. 
Figure 29: Respondent perception of colleagues’ knowledge about the TCC (n=1,346)
3.6.2 Recent services provided to SGBV survivors
The evaluation also sought to measure if Treatment 2 training participants adopt behavior changes in how 
they interact with SGBV survivors after participating in the Integrated Management training. To measure 
this change, respondents were asked at baseline and endline what actions they had taken to assist SGBV 
survivors in the last 60 days. Actions presented in Table 20 include items such as identifying potential 
victims, informing victims about their rights, taking someone to a TCC. As illustrated in the table, we do 
not observe positive changes over time in most of the actions included in the survey. Many of the 
respondents already work with SGBV victims and reported supporting this group at baseline. While the 
54
decrease in many actions is surprising (e.g., identification of victims, receiving a report of SGBV, and 
informing a victim about rights), it is possible that this change is unrelated to the intervention, and we do 
see a positive increase in the two actions directly related to TCCs. More respondents reported taking 
someone to a TCC and informing a victim about TCC resources, the latter of which is statistically 
significant. Interestingly, at baseline, 81 percent reported advising victims of their rights but only 65 percent
reported advising them about TCCs. At endline, however, 74 percent of respondents reported both 
advising victims of their rights and the same percentage reported advising them about the TCCs, suggesting 
that TCCs are now part of the information participants provide to victims. 
 
Table 20: Respondent recent practices with SGBV victims (n=1,346)
In the past 60 days, how often did you…
% 
reporting 
one or 
more 
times at 
baseline
% reporting 
one or 
more times 
at endline
pp 
difference
significant 
at 5%
Identify a student or client that you suspected to be a 
victim of SGBV. 66 55 -11 * 
Have someone report SGBV to you. 72 64 -8 * 
Inform a victim about her/his rights with respect to 
SGBV. 81 74 -7 * 
Coordinate with another service provider to assist a 
victim of SGBV. 68 63 -5 * 
Assess the level of danger a victim of SGBV was 
facing. 61 59 -2 
Document information about a case of SGBV. 63 61 -2 
Contact a service provider on behalf of a victim of 
SGBV. 65 64 -1 
Help a victim of SGBV in a dangerous situation 
establish a safety plan. 58 59 1 
Personally take someone to the TCC to get help. 41 44 3 
Inform a victim of SGBV about other resources. 65 70 5 * 
Inform a victim of SGBV about resources available at 
the TCC 65 74 9 * 
Of those professionals who contacted a service provider on the behalf of a victim, respondents were
seven percentage points less likely to contact the TCC at endline than baseline, and seven percentage 
points less likely to coordinate with the TCCs at endline than baseline. However, respondents were less 
likely to contact other service providers in general at endline, and also less likely to coordinate with 
another service provider at endline, which could potentially be explained by trainees feeling more 
equipped to handle a SGBV case without the assistance of other providers. When we look at the share of 
respondents contacting the TCC compared to other service providers, we see that this share increased 
by three percentage points between baseline and endline. This implies that when a trainee did contact 
another service provider, s/he was slightly more likely to contact the TCC rather than another provider, 
although this relationship is not statistically significant. Similarly, the share of respondents coordinating 
with the TCC compared to other service providers increased by 4 percentage points between baseline 
55
and endline, a statistically significant increase. Respondents were much less likely to contact or coordinate 
with the police at endline.
Table 21: Respondents’ contact and coordination with other service providers (n=857) 
If you contacted a service provider on behalf of a 
victim of SGBV, did you contact…
% Yes at 
baseline
% Yes at 
endline
pp 
difference
PP difference 
in ratio of 
total
TCC 43 36 -7* 3 
Hospital or clinic 43 39 -4 5*
Police 69 42 -27* -7*
Social workers/social development 8 10 1 3*
If you coordinated with another service provider 
on behalf of a victim of SGBV, did you coordinate 
with…
% Yes at 
baseline
% Yes at 
endline
pp 
difference
PP difference 
in ratio of 
total
TCC 43 36 -7* 4*
Hospital or clinic 49 38 -11* 2 
Police 66 44 -22* -2 
Social workers/social development 9 7 -2 0 
Service providers were also asked about their perceived level of difficulty talking with a victim about SGBV
(see Figure 30). Of all the trainees responding to this question, 41 percent stated this is “difficult” or “very 
difficult” at baseline but only 21 percent said so at endline. The number of respondents who said this is 
"easy" or "very easy" increased from 26 percent to 57 percent at endline. These results suggest that the 
training has improved the service providers’ comfort level in talking with victims about SGBV. 
Figure 30: Respondent comfort level talking with victims about SGBV (n=1,290 at baseline and 1,345 
at endline)
56
Respondents were also asked questions about SGBV laws in South Africa and perceptions of SGBV 
reporting and TCC use. At baseline, nearly all respondents (94 percent) knew that South Africa has laws 
that address SGBV; this percentage increased slightly by endline to 97 percent. 
As shown in Figure 31, respondents perceived higher levels of under-reporting of SGBV cases at endline 
and lower incidence of false-reporting. Perceptions about how frequently survivors go to the TCC 
remained relatively unchanged from baseline to endline. 
Figure 31: Perceptions about SGBV reports and TCC usage (n=1,263)
Surveyed trainees perceived higher levels of victim referrals to the TCC at endline than at baseline. The 
share of respondents who thought “all” colleagues were referring victims increased from 23 to 33 percent, 
and response for most colleagues increased from 20 to 32 percent (see Figure 32). 
Figure 32: Perception about number of colleagues referring victims to the TCC (n=1247)
57
3.6.3 Attitudes about SGBV
The Integrated Management training also seeks to influence trainee attitudes toward survivors of SGBV. 
At baseline, many professionals revealed moderate levels of victim-blaming. A third of respondents 
believed that women provoke rape by their appearance or behavior and nearly half indicated that the 
extent of a woman’s resistance should be the major factor in determining if a rape has occurred. This is 
alarming as many rape victims are unable to resist or feel it would be futile to do so. 
The training appears to have improved trainee attitudes in seven of nine categories evaluated (see Table 
22). For example, the share of respondents who believe that the extent of a woman’s resistance should 
be the major factor in determining if a rape has occurred went down from nearly half to a third, and the 
percent who agreed that, “many women claim rape if they have consented to sexual relations but have 
changed their minds afterwards,” dropped by 13 percentage points. Contrary to expectations, agreement 
went down with one statement: "women do not provoke rape by their appearance or behavior." However, 
it is possible that asking respondents to agree with a negative statement created some confusion among 
respondents. 
Table 22: Respondent attitudes toward SGBV (n=1313 at baseline and 1346 at endline)
SGBV Statement
Agree (%) 
at 
baseline
Agree 
(%) at 
endline
pp 
difference
Desired & 
Observed 
Change 
Significance
at 95%
The extent of the woman’s 
resistance should be the major 
factor in determining if a rape has 
occurred.
47 34 -13 ↓↓ * 
Many women claim rape if they 
have consented to sexual relations 
but have changed their minds 
afterwards. 
37 24 -13 ↓↓ * 
Accusations of rape by bar ladies, 
strippers, and prostitutes should 
be viewed with suspicion.
25 18 -7 ↓↓ * 
Many women invent rape stories 
if they learn they are pregnant. 19 15 -4 ↓↓ * 
Women who have had prior 
sexual relationships should not 
complain about rape.
10 7 -3 ↓↓ * 
Women who are raped while 
accepting rides from strangers get 
what they deserve.
6 4 -2 ↓↓ * 
Women often claim rape to 
protect their reputations. 29 30 1 ↓↑ 
Women do not provoke rape by 
their appearance or behavior. 66 55 -11 ↑↓ * 
A raped woman is usually an 
innocent victim. 70 74 4 ↑↑ * 
Professionals were also asked to respond to a series of questions following a hypothetical scenario of 
sexual assault. The purpose of this section was to assess their attitudes toward rape in a potential real￾world scenario. 
Scenario: An attractive 20-year-old single woman wearing a mini-skirt goes out on a Friday night with friends. She 
stays for a few hours and has a few drinks. On her way home, she is assaulted by a man. She is unable to fight 
58
him off and he rapes her. 
Although a minority of responses indicated some victim blaming, at baseline 86% of the respondents either 
disagreed (30 percent) or strongly disagreed (56 percent) that a woman in this scenario is to blame for 
the sexual assault. The percentage of aggregated disagreed/strongly disagreed responses increased to 92
percent after the training, indicating slightly less victim-blaming than at baseline. However, a large number 
of respondents changed their opinions from "strongly disagree" to "disagree,” possibly indicating less 
certainty about the extent to which the woman is to blame than expressed at baseline. 
Figure 33: Respondent level of victim-blaming (n=1,274 at baseline and n=1,340 at endline)
When asked after the training whether this victim would file a police report, respondents were more 
likely to say "sometimes" or "often" instead of "always" (see Figure 34), indicating greater recognition of 
under-reporting of rapes in scenarios similar to this one. Respondents' answers changed in a similar way 
when asked if this victim would go to the TCC, with more respondents now saying "sometimes" or "often" 
instead of "always." These changes in perception might have occurred due to the training, as participants 
now have an increased awareness of the rates of police reports and TCC usage.
Figure 34: Perception of whether victims would file a police report or go to the TCC
59
After the training, fewer respondents think that reporting to the police would be very helpful in this case, 
from 72 percent to 30 percent. Most respondents answered that reporting to the police would be "helpful" 
(see Figure 35).
Figure 35: Perception of police helpfulness (n=1346)
Respondents then identified the main reasons they believed a victim in this scenario might not file a police
report or visit the TCC for help. Fear of being blamed was most commonly identified as one of the main 
reasons for not filing a police report, by 47 percent of respondents. Lack of information was most 
commonly identified as one of the main reasons to not visit the TCC, by 71 percent of respondents. “Staff 
would not be helpful” was the least cited reason in both cases. 
Figure 36: Main reasons for victims not filing a police report or going to the TCC (n=1345)
When asked whether the TCC would be helpful with medical, legal or psychological/emotional assistance, 
fewer respondents answered "not very helpful" or "not at all helpful,", and more respondents said "helpful" 
47%
9%
27%
31%
17%
5%
16%
71%
0%
10%
20%
30%
40%
50%
60%
70%
80%
Fear of being
blamed
Staff would
not be
helpful
Fear of
perpetrator
Lack of
information
Main reasons for not
filing a police report
(% agree)
Main reasons for not
visiting the TCC (%
agree)
60
than at baseline (see Figure 37). The share of professionals in our sample answering "very helpful" declined 
from 75 percent to 47 percent for medical services, from 61 percent to 45 percent for legal assistance, 
and from 78 percent to 47 percent for psychological/emotional assistance.
Figure 37: Perception of TCC helpfulness (n=1345)
3.6.4 Status of action plans
After the training, respondents were asked about their plans to implement SGBV projects in their facilities. 
Most (89 percent) said they planned to implement such a project, and three percent did not yet know. 
When asked whether they recalled the major tasks involved in selecting an appropriate SGBV project for 
their facilities, which were covered during the training, 80 percent of respondents said yes. Of those who 
answered yes, only 85 percent actually explained the process correctly. This means that 68 percent of 
trainees could recall the SGBV project selection process at the time of the survey.
Respondents who said they were planning to implement a SGBV project in their facilities were asked the 
status of project selection. A little more than a quarter (27 percent) of respondents had not yet started 
the process, but only two percent were not planning to do so. One-fifth (22 percent) of respondents had 
already completed their SGBV project selection.
61
Figure 38: Status of SGBV project selection (n=1308)
Those respondents who said that SGBV project selection was complete were then asked about the 
following six activities: expansion of staffing and services; education of the community; re-routing of clients, 
continuity of care and follow-up mechanisms; setting up of referral mechanisms, protocols and policies; 
development of planning and monitoring tools, SGBV material and forms; and education of staff.31 Most 
respondents planned to do all six activities and each was completed at the time of the interview for 22-
26 percent of respondents.
Figure 39: Status of activities within SGBV projects (n=1286) 
 
31 These are the activities covered in the Integrated Management training curriculum. 
27%
22%
17%
22%
2% 1%
0%
5%
10%
15%
20%
25%
30%
Has not
started
Compiling
options
Assessing
options
Selection is
complete
Not planned Don't know
26
23
24
23
22
22
47
43
42
45
59
43
22
28
28
27
15
31
5
6
6
5
3
5
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Education of staff
Development of tools and forms
Referral mechanisms and protocols
Re-routing, continuity of care
Education of the community
Expansion of staffing and services
Status of SGBV project activities
completed in progress not started not planned
62
Respondents who were not planning to implement an SGBV project in their facilities were asked why. 
Figure 40 shows that the main reasons for not planning implementation were lack of time to dedicate to 
the project (24 percent) and a lack of resources (staff, financial, etc.) (23 percent). 
Figure 40: Reasons for not planning an SGBV project at their work facility (n=139)
In summary, findings from the pre- and post-training surveys show that training participants substantially 
increased their knowledge of TCCs and TCC services during the training. Moreover, we observe smaller, 
but statistically significant changes in many attitudes toward SGBV, including less victim-blaming, as well as 
some improvements in reported behaviors in working with survivors of SGBV. Thus, the Treatment 2
training can be regarded as successful in improving the knowledge, attitudes, and to a limited extent, the 
practices of those service providers who attended the trainings. It is important to note that the study did 
not include a control or comparison group for Treatment 2, and, as such, we cannot completely rule out 
the possibility that there are other factors that explain the changes over time. Nonetheless, given that few 
nationwide phenomena could be expected to elicit such consistent change, we feel confident that this is 
not a major methodological concern. Of greater concern, however, is the changing of survey modalities, 
from a self-administered survey to a phone-based survey conducted by an enumerator. It is possible that 
this change in modality influenced responses. It is also important to note that this group of professionals 
self-selected into the training, and as such are not representative of the service providers targeted by the 
training (e.g., teachers, police, and health workers) more broadly. As trainees self-select by choosing to 
participate or not, it is likely that these are individuals who are motivated to learn about TCCs and provide 
assistance to survivors of SGBV. Consequently, these results are not generalizable to all service providers 
in the system of care, and do not necessarily indicate improvements in the system of care as a whole. 
24% 23%
19%
15%
12%
7%
0%
5%
10%
15%
20%
25%
Do not have
the time
Lack of
resources
Other Refused to
answer
Sufficient
projects exist
Lack of will
within my
facility
63
CONCLUSIONS 
• Qualitative and quantitative data collected at TCCs reveal that TCCs vary in capacity and in 
resource availability, with differences in days and hours of operation, appearance and quality of 
facilities, services provided, staffing, recordkeeping practices, and resources. Many TCCs face 
deficiencies in these areas; however, through coordination with NGOs and other stakeholders, 
TCCs are sometimes able to fill service gaps that they could not address on their own, and 
improve the quality of care provided survivors. 
• We find no impact of the community dialogues or the service providers’ training on TCC 
utilization, measured by survivors presenting to the TCC. 
• Predictors of TCC utilization include population size, reported sex crimes, and the murder rate, 
as a proxy for general crime and SGBV. There is also some indication that TCC capacity 
influences utilization; however, the effect of various measures of TCC capacity, such as staffing, 
after-hours operation, facilities, and supplies, is not consistent across the statistical models.
• We also find no impact of the community dialogues on community knowledge and attitudes 
toward sexual assault. Knowledge of TCCs in treatment and control areas remains low and at 
similar levels as baseline. It is possible that the intervention influenced the individuals who 
attended the dialogues; however, such an approach is inadequate to reach the larger population 
and change community knowledge and attitudes.
• The Treatment 2 training can be regarded as successful in improving the knowledge, attitudes, 
and to a limited extent, the self-reported practices of those service providers who attended the 
trainings. It is important to note that this group of professionals self-selected into the training, 
and as such are not representative of the service providers targeted by the training (e.g., 
teachers, police, and health workers) more broadly.
64
RECOMMENDATIONS
Based on the findings and conclusions from this impact evaluation of the ISSSASA activity, the evaluation 
team recommends that USAID, the NPA, and other stakeholders consider the following areas of 
intervention to continue developing the system of care for survivors of sexual assault. 
• Support TCC capacity. TCCs reported various areas of need for support across staffing, 
facilities, and supplies. In a resource-constrained environment and given the relatively low levels 
of TCCs use—where perhaps a TCC only serves one survivor every other day or several times 
a week—justification for added staffing or other resources may be a challenge. On-call staff that 
are available 24 hours a day seven days a week, for example, as opposed to staff sitting at the TCC 
around the clock may be a more reasonable use of resources, especially in areas with low TCC 
use. Continuing to encourage partnerships with NGOs and community organizations, which we 
find to be providing important complementary services and filling resource voids, is another 
promising area for intervention. The findings above indicate some changes that would be effective 
while expending minimal resources. For example, the policy of keeping TCCs’ locations somewhat 
secret undermines utilization and should be reconsidered. • Test the effects of a more broad-based, social marketing approach to raising 
awareness about TCCs and SGBV. Given continued low awareness of the TCCs within 
surrounding communities and continued apprehension about reporting SGBV, more intensive 
outreach efforts should be tested. The broader ISSSASA program includes a television drama 
series, a school based intervention engaging children in grade 7, a radio public service 
announcement campaign, and a digital and social media campaign. Efforts to increase awareness 
and change attitudes should build on best practices in social marketing and be designed to reach 
and influence a wider audience with sustained and repeated messaging via a “saturation” approach. 
Such an approach would be amenable to a follow-on impact evaluation using the same survey 
methodology applied here and could even be focused on communities with the greatest perceived 
need. The intervention would require a stronger theory of change that would be likely to produce 
a change detectable in a random sample of women in targeted communities. As one stakeholder 
noted, “…social change at the community level often occurs gradually and over a more protected
period of time and with repeat exposure as opposed to a one-off exposure to an intervention.” 
Furthermore, given that the TCCs report limited to no direct community outreach activities at 
present, the evaluation team recommends that TCC staff engage directly with communities, 
thereby building connections with schools, clinics, police, community organizations, and 
community members themselves. These direct relationships have the potential to improve the 
status of the system of care, build trust, and increase awareness. Data from this evaluation shows 
that few TCC staff were available to participate in community dialogues or service providers’ 
trainings due to their responsibilities at the TCC or lack of transportation. Human resource 
limitations remain a challenge at the TCCs; however, centralized staff could help coordinate 
outreach efforts. Where it is not possible to send TCC staff into communities for outreach, 
consider technology, social media, or radio as a means for TCC staff to engage with communities. 
• Selectively choose dialogue participants that are potential agents of change. It was 
assumed that messages disseminated in community dialogues would spread throughout 
communities. Findings from this IE show that this assumption did not hold true. If dialogues are 
maintained, they will either need to be scaled up dramatically to reach a critical mass of women 
and men or there should be greater selectivity in who participates in the dialogues. Future 
interventions should consider targeting community members who are potential agents of change 
65
who will proliferate messages from the dialogues. Participants should be provided with the skills 
and materials to encourage such proliferation. 
• Adopt a train-the-trainer approach to community dialogues. Community dialogues are 
an attractive option towards meeting the goals to decrease SGBV incidence and increase support 
for survivors, as they allow for a more in-depth discussion of SGBV issues, which is likely necessary 
to change attitudes and reduce stigma. Nonetheless, these dialogues are only able to reach a small 
number of participants, and we did not find evidence that they were effective in achieving the 
objectives tested in this evaluation. As such, they should either be used strategically as part of a 
“saturation” strategy in targeted communities, or they should be scaled-up dramatically. As part 
of the latter approach, we would recommend adopting a train-the-trainer or promoter-based 
approach. There is some precedent for this, as Soul City has been working to establish clubs of 
young women who could lead their own dialogues. Participants should be provided with the skills 
and materials to encourage information proliferation while ensuring fidelity to the established 
approach. 
• Formalize SGBV training for police and other professionals in the system of care. 
Participants in the Integrated Management training were recruited via open invitations to service 
providers at targeted institutions. As such, participants self-selected into the program and many 
had a strong interest in SGBV or were already working on SGBV. In fact, at baseline, 80 percent
of service providers already knew of SGBV services and 65 percent had informed a victim about 
the TCCs in the last 60 days. This could explain why the intervention did not increase referrals 
from this group. It seems probable that a training could be most beneficial if it were provided to 
service providers who likely have contact with SGBV survivors but are not knowledgeable about 
such issues nor the services that exist to address them. These individuals would be less motivated 
to attend a three-day training, but a shortened one-day training could be offered to a wider 
audience of service providers with a heavy emphasis on follow-up action. 
• Police would be an obvious target for such training; however, USAID/South Africa reports that it 
is difficult to work with the South Africa Police Service in such capacity due to Leahy Law 
requirements for training. Training nurses in making referrals would be an attractive alternate
target group, as these professionals often interact with survivors. While nurses have many 
demands on their time, they might be amenable to such a training because the TCCs offer a 
support service that can make their jobs easier. If this were to occur, we would recommend 
conducting the same pre- and post-survey. Some of the knowledge and attitude gains might be 
lower with a less motivated group; however, a well delivered training could lead to greater 
awareness throughout the service provider community and increase referral sources.
• Use statistics in this report to inform future programming: The survey data demonstrate 
a number of knowledge gaps, important perspectives, and potentially problematic attitudes that 
could help USAID, the NPA, and the IPs improve their programming. For example, while survey 
data reveals generally progressive attitudes, it also shows that many individuals think there are 
limits on a woman’s right to refuse sex and there is evidence of victim blaming in certain situations. 
These statistics should be incorporated into curriculum, discussed, and addressed. 
66
ANNEXES
ANNEX I: EVALUATION STATEMENT OF WORK32
Impact Evaluation Strategy for South Africa GBV Project 
Eric Mvukiyehe 
November 15, 2012
Background 
Over the past five years, USAID South Africa has worked with the South African Government and other 
development partners to set up dozens of Thuthuzela Care Centers (TCCs), one-stop facilities designed 
to provide clinical services and psychological counseling to victims of rape and other sexual assaults. 
Currently, there are 52 TCCs around the country, at least one in each of the nine provinces. The Mission 
is now launching a new five-year project titled “Increasing Services for Survivors of Sexual Assault in South 
Africa,” which will be implemented by the Foundation for Professional Development (FPD) along with Soul 
City and Soke as sub-implementers.
This new project seeks to move beyond service provision to also focus on broader issues of Gender￾Based Violence (GBV) prevention through a wide range of public awareness-raising and capacity￾strengthening activities. Specific objectives of the project are: (i) to increase public awareness of the 
services provided at all 52 TCCs; and (ii) to expand and improve the services provided at TCCs and in 
the TCC catchment areas. If met, these objectives should contribute to increase in utilization TCC clinical 
service by the survivors of rape and other sexual assaults and more generally to a change in attitudes and 
behaviors about rape as well as to a decrease in risks and incidence of GBV in TCC-catchment areas. 
An impact evaluation study is being explored to ascertain the effectiveness of this project. 
This memo outlines the broader contours of design options for such a study.
Evaluation objectives
Based on the aforementioned project objectives and on conversations with the stakeholders, 
the following evaluation objectives are achievable: 
• To ascertain the effects of project activities on survivors’ propensity to seek out and utilize TCC 
services 
• To ascertain the effects of project activities on survivors’ psychological and social wellbeing 
• To ascertain the effects on GBV-related attitudes and behaviors on the part of community members, 
healthcare professionals and law enforcement authorities 
• To ascertain the effects of project activities on incidence of GBV in communities 
Activities to be evaluated 
 
32 While the USAID approved design is based on this SOW, some aspects of the SOW were changed or dropped in the approved 
design. For example, the evaluation does not answer if capacity strengthening activities improve survivors’ psychological and social 
wellbeing. 
67
This research study will focus on the efficacy of different project activities on key outcomes of interested. 
Thus far, it seems that the main outcomes have to do with (i) increasing survivors’ propensity to seek out 
TCC services and staying through the counseling and legal processes; and (ii) changing attitudes and 
behaviors on the part of members of the broader community, including survivors’ families and 
professionals involved in the referral systems. Presumably, project implementers have identified the list of 
key factors that contribute to the problems that underlie underutilization of TCC services and prevalence 
of GBV such that the proposed activities are designed to address these root causes. From what I can 
gather in the project documents and the conversations with implementing partners, the following barriers 
or risk factors have been singled out:
• Lack of information about availability (and benefits) of TCC services
• Social structures that embed permissive gender social norms (e.g., patriarchy) and condone rape 
and other sexual assaults or stigmatize the victims
• Weak protection environments and lack of trust in institutions and services that are supposed 
provide assistance to survivors (e.g., police, TCC and hospital staff, law enforcement services, 
etc.)
Thus, it seems to me that the array of project activities designed to address these risk factors fit in two 
broad categories: (i) public awareness-raising activities aiming to provide information about TCC 
services and to educate the public about GBV issues; and (ii) capacity-strengthening activities aiming 
to enhance service delivery and/or to create a safe and trusting environment for the survivors. Arguably, 
some activities such as “open days” may have a dual purpose.
Target populations of interest
This project targets at least three primary populations of interest:
• Survivors of rape and other sexual assaults
• Survivors’ families and fellow community members (in TCC catchment areas)33
• Professionals who are part of the referral systems (e.g., TCC and hospital staff; police; 
prosecutors; local NGOs; etc.)
Key outcome areas of interest 
Outcomes of interest are indicators of change, which can tell you whether project activities have been 
effective or not and are typically operationalized from project and evaluation objectives. These indicators 
have to be measurable empirically, either through surveys or some other ways. One way to organize these 
outcomes is to thinking about the main project activities that will be carried out and the population of 
interest these activities will be targeting and ask yourself the following question: “what changes should I 
expect to see on this population of interest if project activities are effective?” From this perspective, it seems 
that there are three primary indicators of change and a number of secondary ones.
Primary outcome #1: Survivors’ propensity to seek out and utilize TCC services
• Knowledge about TCC services and benefits 
 
33 This category encompasses many different subgroups, including, potential victims of rape and other sexual assaults (e.g., girls 
and women at risk of GBV); potential perpetrators of GBV crimes; victims’ families; ordinary community members; community 
leaders; among others. 
68
• Seek out treatment 
• Satisfaction with TCC services 
• Staying on through a counseling plan 
• Trust in referral systems
• Willingness to report GBV crimes
• Likelihood to stay through the legal processes or to withdrawal case
• Psychological and social wellbeing (e.g., less-strained relations with family community members; 
positive outlook; paranoia; sense of self-worth; etc.)
• Civic engagement 
Primary outcome #2: Change attitudes and behaviors about GBV issues on the part of referral 
systems professionals as well as the survivor’s family and community 
• Knowledge about GBV issues
• Attitudes about GBV issues 
• Empathy and support toward survivors
• Behavioral intent regarding GBV
• Attitudes toward women and gender rights
Primary outcome #3: Decrease in risk of exposure to (and in incidence of) of GBV
• Prevalence of GBV-risk factors 
• Prevalence of GBV 
Research questions
The research questions will gauge the extent to which specific project activities (or combination thereof) 
influence the outcomes of interest on a given target population. The following are suggested generic 
questions, which can be refined based on the theories of change that underlie specific activities. 
1. Do awareness-raising activities increase the likelihood that survivors will seek out TCC 
services and go through the entire counseling and legal process? 
2. Do capacity strengthening activities improve survivors’ psychological and social wellbeing?
3. Do awareness-raising activities lead to changes in attitudes and behaviors about GBV?
4. Do awareness-raising activities lower the risk of exposure to (and incidence of) GBV? 
5. Do capacity strengthening activities increase the likelihood that survivors will seek out TCC 
services and stay through counseling and legal processes?
6. Do capacity strengthening activities improve survivors’ psychological and social wellbeing?
7. Do capacity strengthening activities lead to changes in attitudes and behaviors about GBV?
8. Do capacity strengthening activities lower the risk of exposure to (and incidence of) GBV? 
69
Hypotheses
Hypotheses are conjunctures between specific project activities and the key outcomes of interest, based 
on the program’s theory of change. That is, these are provisional answers to your research questions, 
pending confirmation from empirical evidence. To construct sound hypotheses, you need to ask yourself 
the following questions: “Which project activities are likely to produce the desired change on a particular 
population of interest? Will such change occur under any circumstances or will change depend on other factors?” 
[NOTE TO IMPLEMENTING PARTNERS: IT WOULD BE HELPFUL IF YOU COULD COME UP 
WITH A LIST OF THE MAIN ACTIVITIES YOU PLAN TO CARRY OUT, THE POPULATION WILL BE 
TARGETED AND THE KIND OF CHANGE YOU EXPECT TO SEE ON THIS POPULATION AND 
WHY AT THE END OF THE INTERVENTION.]
Based on our conversations, I propose the following hypotheses, but these will have to be refined or 
modified depending on your precise understanding of the theories of change on which your interventions 
rest. Please do keep in mind that we have different populations and outcomes of interest and so project 
activities may not have the same effects on these. Thus, in refining these hypotheses, we will be paying 
attention to such potential differences.
Hypotheses about survivors’ attitudes and behaviors 
This population of interest is targeted by both public awareness raising (whether though radio programs 
or community dialogue) and capacity-strengthening activities, though at different stages of the process. 
Thus, it could be argued that public awareness-raising activities may increase the likelihood that an 
individual survivor seeks out TCC services, but we wouldn’t expect that public awareness alone will 
influence the likelihood that these individuals stay on course of counseling program and through the legal 
process. The latter outcomes will probably depend on whether this individual trusts the systems and feels 
safe enough, which in turn depends on professionals getting training to create a safe environment for 
survivors. Thus, the following hypotheses can be formulated: 
H1. Awareness-raising activities (e.g., community dialogues; radio programs) will increase survivors’ knowledge 
about TCC services and increase survivors’ likelihood to seek out TCC services
H2. Capacity-strengthening activities (e.g., multidisciplinary training) will increase the likelihood that survivors will 
stay through the process (counseling and legal) 
H3. Capacity-strengthening activities (e.g., multidisciplinary training) will improve survivors’ psychological and social 
well-being
H4. Capacity-strengthening activities (e.g., multidisciplinary training) will improve survivors’ psychological and social 
well-being only if survivors have also been exposed to GBV awareness-raising activities 
Hypotheses about attitudes and behaviors of referral systems professionals 
This population interest encompasses a variety of groups including TCC and hospital staff; law 
enforcement authorities (e.g., police; prosecutors); members of local NGOs, among others. Arguably, this 
population is primarily targeted through capacity-strengthening activities, even though public awareness 
activities can also have some influence indirectly. Thus, the following hypotheses are suggested:
H5. Capacity-strengthening activities (e.g., multidisciplinary training) will change the way referral systems 
70
professionals provide assistance to survivors
H6. Capacity-strengthening activities (e.g., multidisciplinary training) will change GBV attitudes and behaviors on 
the part of referral systems professionals 
H7. Capacity-strengthening activities (e.g., multidisciplinary training) will change GBV attitudes and behaviors on 
the part of referral systems professionals only if these professionals have also been exposed to GBV awareness￾raising activities
Hypotheses about attitudes and behaviors of survivors’ families and of fellow community 
members 
This population of interest is probably targeted primary through public awareness activities (e.g., 
community dialogues; radio programs; open days) and we wouldn’t expect capacity strengthening activities 
to influence its attitudes and behaviors directly. However, there could be some indirect influence. For 
example, if policy or prosecutors who receive training change the way they handle GBV cases then would￾be perpetrators will take this into account before they engage in crime. Thus, I propose the following 
hypotheses:
H8. Awareness-raising activities (e.g., community dialogues; radio programs) will change GBV attitudes and 
behaviors of community members 
H9. Awareness-raising activities (e.g., community dialogues; radio programs) will decrease the risk of exposure to 
(and incidence of) GBV in communities 
H10. Capacity-strengthening activities (e.g., multidisciplinary training) will decrease the risk of exposure to (and 
incidence of) GBV in communities 
H11. Capacity-strengthening activities (e.g., multidisciplinary training) will decrease the risk of exposure to (and 
incidence of) GBV in communities only if community members have also been exposed to awareness-raising 
activities
H12. The effects of either awareness-raising activities or capacity-strengthening activities on community members’ 
attitudes and behaviors about GBV will depend on socioeconomic conditions that prevail in each community
Identification of effects of project activities
Programmatic and logistical constraints make it difficult to carry out program activities across all TCC 
areas simultaneously. Therefore, we propose to randomize targeted areas in two groups, whereby the 
first group will receive some types of project activities (either awareness-raising or capacity-building or 
both) as soon as the project launches, the other group will receive project activities in the second wave. 
A phase-in strategy would be followed to identify the effects of project activities on key outcomes for 
each of the three populations of interest. First, the project period will be divided in two waves, whereby 
in the first part of project life different intervention options are tried out in different communities and in 
the latter part of project life the best performing intervention(s) get(s) rollout in all targeted communities. 
As activities in the first wave are phased out and before activities in the second wave are phased in, we 
will gather follow-up data on both groups to ascertain the effects of project activities. These results can 
inform which activities are the most effective and perhaps focus on those going forward. I provide details 
below. How to divide the two waves is subject to discussion. One suggestion would be as follows: 
71
• Wave#1: First 12-20 months of project life 
• Follow-on data collection: 2-4 months after wave one is completed
• Wave #2: Remaining projects life (i.e. the last 36 months)
[NOTE TO IMPLEMENTING PARTNERS: NOTHING IS SET HERE. OTHER SCENARIOS ARE 
FEASIBLE. KEY IS TO SELECT A PERIOD FOR WAVE ONE THAT YOU THINK WILL BE SUFFICIENT 
FOR PROJECT ACTIVITIES TO HAVE MEANINGFUL EFFECTS.] 
This strategy will enable to ascertain project effects in two ways. 
Ascertaining the efficacy of any type of project activities 
The first approach would be to compare outcomes of interest between communities that receive some 
type of project activities and those that haven’t yet. As suggested above, target communities or individuals 
will be divided in two groups, a larger subset of communities or individuals that receive different types of 
project activities in wave#1 and a smaller subset of where project activities are delayed until wave#2. 
[NOTE: WE ARE NOT TALKING ABOUT CLINICAL SERVICES HERE. THAT ASPECT OF THE 
PROJECT IS UNTACHED.] The idea is that at the end of the first wave (and before rollout of the second 
wave), we will gather from both groups and ascertain program effectiveness by comparing the outcomes 
of interest between the two groups—that the groups that did and did not receive any program activities 
in the first wave.
Ascertaining the efficacy different types of project activities 
The second approach will allow us to ascertain not only whether a program works or not, but also what 
specific program activities have the greater effects on the key outcomes of interest. Thus, within the first 
wave, target communities or individual beneficiaries will receive different types of project activities: some 
will receive awareness-raising activities only, other will receive capacity-strengthening activities only and 
others will receive some combination of both. Thus, a key aspect of the evaluation here is to investigate 
which of the different intervention options has the greatest impact on key outcomes of interest for the 
different target populations.
Sampling plan 
Sampling here is tricky. The difficulty is that the structure of the community is not very well defined, partly 
as a result of TCCs’ emplacement in populated areas that are most at risk of GBV. Another difficultly is 
that this study is targeting two different populations of interest (i.e. victims of rape and other sexual 
assaults and community member) whose distribution may not completely overlap. With these problems 
in mind, I suggest a design that presumes that we are interested in learning about all three populations of 
interest (i.e., survivors; referral systems professionals and survivors’ families and communities). 
Use ‘community’ as the primary unit of treatment for survivors and fellow community 
members 
One option would be to use “community” as a unit of treatment, meaning that project activities will be 
assigned to different communities and assume that all individuals within a particular community have been 
exposed to such activities. The difficulty here, though, is that the study will require a lot more communities 
that we may have initially wanted to work in. This is because the study requires a minimum number of 
units required to be able to detect the effects of project activities, if they do exist. 
72
Based on power calculations, we determine that a minimum number of 208 communities will be required 
to adequately detect meaningful effects of project activities of any types, if they exist, and to be able to 
distinguish the relative efficacy of four different kinds of project activities.34 One way to go about drawing 
this sample is to select four communities in TCC’s catchment areas that are equally eligible to receive 
project activities. Three-fourths of these communities (i.e. 156) will receive different types of project 
activities (e.g., awareness-raising, capacity-building or both) in the first wave of project rollout, while the 
remaining quarter will receive activities in the second wave. The main thing here is that that 
community goes through the program first and what types of activities they receive will be 
decided through lottery (public or private). 
In terms of data collection, we would not need to interview every single individual in these communities. 
Rather, we can select a sample as small as 30 individuals (10 survivors; 10 professionals and 10 fellow 
community members) and as big as 60 individuals, depending the level of detailed subgroup analysis we 
want to investigate. Table 1 below provides an illustration of one possible design option.
Table 1. Impact Evaluation Design
There are a number of difficulties to keep in mind, however. One big unknown is the structure of 
communities in different catchment areas. We know that TCCs that are in rural provinces are likely to 
be surrounded by more organic, structured and distinguishable villages, but the same is probably not true 
of TCCs located in urban areas where there is likely to be undifferentiated townships and informal 
settlements. For the purpose of the study, we will need to define what we consider “community” in TCC 
 
34 A review of prior studies of sexual assault education programs suggests an average effect size of .30 to .35, meaning that typically 
those attending a sexual education program tend to have about a third of a standard deviation better than those who do not 
(Anderson and Whiston 2005). However, many of these studies have been carried out in the US context and we allow for the 
possibility that GBV programming in developing countries may have much smaller effect sizes. 
Communities in Wave #1 
(Y1 & 2)
Wave #2 (Y2-5)
Intervention type #1
Community dialogue
(26 communities) 
Intervention type #2 
Community radio
(26 communities) 
Implement
Intervention type #3 
Multi-disciplinary training 
(26 communities)
Most successful
Intervention type #4 
Multi-disciplinary training plus 
(26 communities)
Intervention 
Intervention type #5
Combination of public 
awareness-raising and capacity￾strengthening 
(26 communities)
In all 208 communities
Intervention type #6 
Pure control in wave#1
(26 communities)
73
catchment areas, which should be eligible for project activities and ensure that we have enough of these 
in each area to make this design option work. 
Another concern is that the distribution of some of population of interests, especially survivors and 
referral systems professionals, may be uneven in different TCC catchment areas. For example, it is 
plausible that some communities may not have any members from these categories, while others may 
have an over-representation. In addition, in some cases, survivors who use services at a particular TCC 
may be coming from distant communities, rather than from TCC catchment areas. Thus, possible lack of 
adequate overlaps between the three populations of interest is going to be challenge for both targeting 
project activities to the relevant populations of interest and for conducting sampling for the baseline and 
follow-up data collection. One alternative strategy to keep in mind is that these two categories may be 
sampled at the individual, rather than community level. But in that case we would need to think carefully 
about the mechanisms through which individuals are expected to get exposure to the different project 
activities. 
74
ANNEX II: DATA COLLECTION INSTRUMENTS 
Service Provider Survey
Baseline Professional Survey Questionnaire
Thank you for agreeing to participate in the survey. We very much appreciate your help. Please take the time 
to answer the following questions as accurately as you can. There are no right or wrong answers. We just 
want to learn the opinion of trainees on the following issues. All answers will be confidential and will not be 
identified with your name.
Section 0: Introduction
A1. In what Police Precinct do you work?
 ________________________________________ 
A2. What is the name of the TCC in your area? 
_________________________________________ 
A3. What is your Profession? 
[ ] Health Worker 
[ ] TCC Site Coordinator
[ ] Victim Assistance Officer
[ ] Police officer
[ ] NGO Worker
[ ] Educator
[ ] Social Worker 
[ ] Other
A3a. If you marked other, please specify:
________________________________________ 
Section A: Thuthuzela Care Centres
A4. Do you know what services the Thuthuzela Care Centres offer?
[ ] Yes 
[ ] No 
A5. Are there services available in your community for victims of sexual assault? 
[ ] Yes 
[ ] No 
[ ] Don’t Know
A6. Of the following services, which are offered by Thuthuzela Care Centres? (PLEASE SELECT 
ALL THAT APPLY)
[ ] Transportation 
[ ] Medical assistance
[ ] Psychological and counseling assistance
[ ] Legal assistance
[ ] Don’t Know
A7. Do you know the location of the nearest Thuthuzela Care Centre (a help Centre for victims of 
sexual violence)? 
[ ] Yes 
[ ] No 
[ ] Don’t Know
A7a. If you answered yes to the previous question please state the location
_________________________________________ 
75
A8. How many of your colleagues in your workplace (e.g., school, police station, office) know about 
the Thuthuzela Care Centre?
[ ] None
[ ] A few
[ ] Many 
[ ] Most 
[ ] All 
[ ] Don’t Know
A9. Who can go to the Thuthuzela Care Centre for help? (PLEASE SELECT ALL THAT 
APPLY)
[ ] Women 
[ ] Men
[ ] Girls
[ ] Boys
[ ] Lesbian, gay, bisexual, and transgender
[ ] Don’t Know
A10. Can a person under age 18 receive help from a Thuthuzela Care Centre?
[ ] Yes 
[ ] No 
[ ] Don’t Know
A11. Can a man receive help from the Thuthuzela Care Centre if he has experienced sexual violence?
[ ] Yes 
[ ] No 
[ ] Don’t Know
A12. How often do those who go to the Thuthuzela Care Centre have to pay for the services?
[ ] Always
[ ] Often
[ ] Sometimes
[ ] Rarely 
[ ] Never
[ ] Don’t Know
A13. If a person goes to a Thuthuzela Care Centre for help, is she/he required to report the name of 
the person who attacked her/him?
[ ] Yes 
[ ] No 
[ ] Don’t Know
 
A14. Is it possible to file a police report at the Thuthuzela Care Centre without having to go to the 
police station?
[ ] Yes 
[ ] No 
[ ] Don’t Know
A15. If a person goes to the Thuthuzela Care Centre for help, is she/he required to prosecute or take 
legal action against the person who attacked her/him?
[ ] Yes 
[ ] No 
[ ] Don’t Know
A16. How many of your colleagues at your workplace know about the Thuthuzela Care Centre 
services?
[ ] None
[ ] A few
[ ] Many 
[ ] Most 
[ ] All 
[ ] Don’t Know
A17. How many of your colleagues at your workplace regularly refer victims to the Thuthuzela Care 
Centre?
[ ] None
[ ] A few
[ ] Many 
[ ] Most
[ ] All 
[ ] Don’t Know
Section B: Recent Practices
Staff can respond to victims of violence in many ways. In the past 60 days, how often did you do each of 
the following on average? 
B1. Had someone report sexual or gender-based violence to you.
[ ] Never [ ] 2-4 Times
76
[ ] Once [ ] 3-5 Times
B2. Identified a student or client that you suspected to be a victim of sexual or gender-based violence.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B3. Documented information about a case of sexual or gender-based violence. 
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B4. Informed a victim about her/his rights with respect to sexual or gender-based violence.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B5. Informed a victim of sexual or gender-based violence about resources available at the Thuthuzela 
Care Centre.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B6. Personally taken someone to the Thuthuzela Care Centre to get help.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B7. Informed a victim of sexual or gender-based violence about other services (please specify):
[ ] Never 
[ ] Once 
[ ] 2-4 Times 
[ ] 3-5 Times  
B7a. If you informed a victim about other services please specify what service. 
__________________________________
B8. Assessed the level of danger a victim of sexual or gender-based violence was facing.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B9. Helped a victim of sexual or gender-based violence in a dangerous situation establish a safety plan.
[ ] Never 
[ ] Once
[ ] 2-4 Times
[ ] 3-5 Times
B10. Contacted a service provider on behalf of a victim of sexual or gender-based violence. 
[ ] Never 
[ ] Once 
[ ] 2-4 Times 
[ ] 3-5 Times 
B11. If you have contacted a service provider on behalf of victim, which type(s) of service provider did 
you contact? (PLEASE SELECT ALL THAT APPLY)
[ ] TCC
[ ] Hospital
[ ] Police 
[ ] Other 
B11a. If you selected other, please specify
________________________________________
B12. Coordinated with another service provider to assist a victim of sexual or gender-based violence
[ ] Never 
[ ] Once 
[ ] 2-4 Times 
[ ] 3-5 Times 
B13. If you have coordinated with another service provider, which type(s) of service provider did 
you coordinate with? (PLEASE SELECT ALL THAT APPLY)
[ ] TCC
[ ] Hospital
[ ] Police 
[ ] Other
B13a. If you selected other, please 
specify________________________________________
B14. How difficult or easy is it for you to talk with a victim about sexual or gender-based violence?
[ ] Very difficult 
[ ] Difficult
[ ] Neither difficult nor easy
[ ] Easy 
[ ] Very easy 
77
Section C: Sexual Assault and the Legal System
Now please consider a few questions about laws in South Africa.
C1. Are there any laws in South Africa that address sexual and gender-based violence?
[ ] Yes
[ ] No
[ ] Don’t Know
C2. What percentage of sexual or gender-based violence cases do you think are actually reported to 
the police? 
[ ] 0-25%
[ ] 26-50% 
[ ] 51-75% 
[ ] 76-100%
[ ] Don’t Know
C3. What percentage of sexual or gender-based violence cases that are reported do you think are 
false (the person reporting was not actually assaulted)?
[ ] 0-25%
[ ] 26-50% 
[ ] 51-75%
[ ] 76-100%
[ ] Don’t Know
C4. What percentage of sexual or gender-based violence victims do you think go to the TCC for 
treatment or support?
[ ] 0-25%
[ ] 26-50% 
[ ] 51-75%
[ ] 76-100%
[ ] Don’t Know
C5. Have you ever received training related to sexual or gender-based violence?
[ ] Yes 
[ ] No
[ ] Don’t Know 
Section D: Perceptions
This section will ask you about your views regarding various issues in society. We are interested in your views 
regarding these statements. Please feel free to answer any way you like -- there are no right or wrong answers. 
For each statement, please state whether you strongly agree, agree, disagree or strongly disagree with each 
statement.
D1. The extent of the woman’s resistance should be the major factor in determining if a rape has 
occurred.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D2. A raped woman is usually an innocent victim. 
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D3. Women often claim rape to protect their reputations.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D4. Women who have had prior sexual relationships should not complain about rape.
[ ] Strongly agree
[ ] Agree
[ ] Strongly disagree
[ ] Don’t know
78
[ ] Disagree
D5. Women do not provoke rape by their appearance or behaviour. 
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D6. Women who are raped while accepting rides from strangers get what they deserve.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D7. Many women invent rape stories if they learn they are pregnant.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D8. Many women claim rape if they have consented to sexual relations but have changed their minds 
afterwards.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
D9. Accusations of rape by bar ladies, strippers, and prostitutes should be viewed with suspicion.
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
Section E: Case Scenario 
An attractive 20-year-old single woman wearing a mini-skirt goes out on a Friday night with friends. 
She stays for a few hours and has a few drinks. On her way home, she is assaulted by a man. She is 
unable to fight him off and he rapes her. 
E1. Would you agree or disagree with the following statement: “Because of this woman’s behaviour, 
she was to blame for being sexually assaulted.”
[ ] Strongly agree
[ ] Agree
[ ] Disagree
[ ] Strongly disagree
[ ] Don’t know
E2. For women in this neighbourhood who experience such situations, would they file a police 
report? 
[ ] Always
[ ] Often
[ ] Sometimes
[ ] Rarely 
[ ] Never 
E3. Would it be helpful for this woman to report this situation to the police?
[ ] Very helpful
[ ] Helpful 
[ ] Not very helpful
[ ] Not at all helpful
E4. Among the following reasons, what would be the main reason that most women would not file a 
police report in this case? 
[ ] Fear of being blamed
[ ] The police staff would not be helpful
[ ] Fear of the perpetrator
[ ] Lack of information about where to find 
help
[ ] Other
E4a: If other, please specify
___________________________
79
E5. Considering women in this neighbourhood who experience such situations: would they go to the 
Thuthuzela Care Centre? 
[ ] Always 
[ ] Often
[ ] Sometimes 
[ ] Rarely 
[ ] Never
E6. Among the following reasons, what would be the main reason that a woman would not go to a 
Care Centre in this case? 
[ ] Fear of being blamed
[ ] The Centre staff would not be helpful 
[ ] Fear of the perpetrator
[ ] Lack of information about where to find 
help
[ ] Don’t Know
E7. Would a Thuthuzela Care Centre be helpful in this situation in any of the following ways?: 
Medical
[ ] Very helpful
[ ] Helpful
[ ] Not very helpful 
[ ] Not at all
E8. Would a Thuthuzela Care Centre be helpful in this situation in any of the following ways?: Legal
[ ] Very helpful
[ ] Helpful
[ ] Not very helpful 
[ ] Not at all
E9. Would a Thuthuzela Care Centre be helpful in this situation in any of the following ways?: 
Psychological/emotional
[ ] Very helpful
[ ] Helpful
[ ] Not very helpful 
[ ] Not at all
E10. Thinking about what would happen in such cases: assume 10 such police reports were filed this 
year, how many of these reports would ultimately result in prosecution? [Your best guess is fine.]
______________________________
Section F: Background
Now I am going to ask a few final questions about you.
F1. What is your gender?
[ ] Female
[ ] Male
[ ] Other
F2. What is your age?
[ ] Younger than 18
[ ] 18-30 years old
[ ] 31-40 years old
[ ] 41-50 years old
[ ] 51-60 years old
[ ] over 60 years old 
F3. Have you attended a previous training on sexual assault?
[ ] Yes 
[ ] No
[ ] No
F3a: If you answered yes to the previous question, please specify to date of the most recent 
training on sexual assault that you attended.
__________________________________ __________________
 (Month) (Year)
Section G: Contact Information 
80
We plan to contact you again three months after this training to follow-up. Please indicate the best 
phone number to reach you, and an alternate phone number. If you have an email address, please list 
that as well.
G1. Phone number (primary):
______________________________
G2. Phone number (alternative):
______________________________
G3. Email:
______________________________
81
Women’s Survey
Instructions: Thank you for agreeing to participate in this study. We really appreciate your assistance. This 
survey will take approximately 30 minutes to complete. Please take the time to answer the following 
questions as accurately as you can. There are no right or wrong answers, we just want to find out what 
people in this community know about sexual violence and related issues. All answers will be confidential 
and will not be identified with your name. You are under no obligation to complete the survey, and are 
welcome to leave out any of the questions that you do not want to answer. 
A.0) Below are a list of crime problems. For each, please state whether it is: not a problem, a minor 
problem, or a major problem in your community. 
1. Domestic 
violence
0=not a problem, 1= a minor problem, 2=a problem, 3=a major problem
2. Gang violence 0=not a problem, 1= a minor problem, 2=a problem, 3=a major problem
3. Sexual assault 0=not a problem, 1= a minor problem, 2=a problem, 3=a major problem
4. Mugging 0=not a problem, 1= a minor problem, 2=a problem, 3=a major problem
5. House breaking 0=not a problem, 1= a minor problem, 2=a problem, 3=a major problem
A.1) Knowledge
Question Response
Read: Sometimes women and men-or even girls or boys- are sexually assaulted 
without their consent. This could include rape, any unwanted sexual contact, or 
being forced or threatened into any unwanted sexual acts. I am now going to ask 
you some questions about your knowledge of help that is available for people who 
have been victims of such sexual assaults in this community.
6. Are there services available in your community for victims of sexual 
assault? 
 [If no/dk/rf, skip to Q10]
1=Yes 2=No
3=Don’t know 4=rf
7. If yes, who provides these services?
[Do not prompt, enumerator codes all categories that respondent mention if a victim 
mentions the applicable NGO name in that region but does not specifically call it an 
NGO, still code this as “NGO”s]
a) Hospital/clinic 
(TCC/NGO 
not 
mentioned)
b) NGO Help 
desk
c) Police
d) TCC(even if 
within a 
hospital/clinic)
e) NGO (even if 
within a 
hospital/clinic)
f) Other: 
__________
___
8. If yes, please tell me what types of support services are provided? You 
can provide more than one answer.
[Do not prompt, enumerator codes all categories that respondent mentions]
g) Medical
h) Legal 
i) Psychological/
emotional
82
j) Spiritual/religi
ous 
Other:__________
_______ 
9. If you were a victim of a sexual assault, who would you first report to 
in order to get help? [Do not prompt, code only one]
a) a close friend 
or neighbour
b) a family
member
c) police
d) hospital or 
clinic 
(TCC/NGO 
not 
mentioned)
e) TCC (even if 
located within 
hospital/clinic)
f) NGO (even if 
located within 
hospital/clinic, 
if applicable)
g) Other 
h) No one
10. In the past three months, have you heard or seen any advertisements, 
announcements, or spots promoting awareness of sexual assault?
1=Yes 2=No 
3=Don’t know 
4=rf 
11. In the past three months, have you heard or seen any advertisements, 
announcements, or spots promoting awareness of resources available 
to victims of sexual assault?
1=Yes 2=No 
3=Don’t know 
4=rf 
12. If yes, what were the main messages do your remember? 
[Do not prompt, enumerator codes all categories that respondent mentions]
1=Awareness of 
sexual violence 
and sexual assault 
as a problem
2=Where to seek 
help
3=Rights of 
victims
4=Messages to 
deter 
perpetrators
5=Other: 
_____________
__
6=dk
83
7=rf
13. Have you heard of the Thuthuzela Care Centres? 
[If yes, skip to 16]
1=Yes 2=No
3=Don’t know 4=rf
Read: A Thuthuzela Care Centre is a crisis centre, to help victims of sexual assault. 
14. Do you know of any places like this in your community?
[If no, Skip to 30}
1=Yes 2=No
3=Don’t know 4=rf
Read: As you may know, a Thuthuzela Care Centre is a crisis centre, to help victims 
of sexual assault. 
15. Do you know what services the Thuthuzela Care Centres offer? 
1=Yes 2=No
3=Don’t know 4=rf
16. Do you know the location of the nearest Thuthuzela Care Centre? 1=Yes 2=No 
3=Don’t know 
4=rf 
Location:
17. Of the following services, which are offered by the Thuthuzela care 
Centres? 
a) Transportation
b) Medical assistance
c) Psychological and counseling assistance
d) Legal assistance
a) 1=Yes 2=No 
3=dn 
b) 1=Yes 2=No 
3=dn 
c) 1=Yes 2=No 
3=dn 
d) 1=Yes 2=No 
3=dn
18. Can a man receive help from the Thuthuzela Care Centre if he has 
experienced sexual assault? 
1=Yes 2=No 
3=don’t know 
4=rf
19. Can a person under age 18 receive help from the Thuthuzela Care 
Centre, even without parental consent? 
1=Yes 2=No 
3=don’t know 
4=rf
20. Do those who go to the Thuthuzela Care Centre have to pay for the 
services? 
1=Always
2=Often 
3=Sometimes
4=Rarely
5=Never
21. If a person goes to the Thuthuzela Care Centre for help, is she/he 
required to report the name of the person who attacked her/him? 
1=yes
2=no 
3=dk 
4=rf
22. Is it possible to file a police report at the Thuthuzela Care Centre 
without having to go to the police station?
1=Yes
2=No 
3=Dk 
4=rf
23. If a person goes to the Thuthuzela Care Centre for help, is she/he 
required to prosecute or take legal action against the person who 
attacked her/him? 
1=Yes
2=No 
3=Dk 
4=rf
84
24. In thinking of the adult women in your neighbourhood, how many of 
them do you think know about the Thuthuzela Care Centre?
0=None
1=A few
2=Some
3=Most
4=All
5=Dk 
6=rf 
25. How many women (18 and older) do you know personally that have 
visited a Thuthuzela Care Centre (or were taken there) to get help? 
26. How many men (18 and older) do you know personally that have 
visited a Thuthuzela Care Centre (or were taken there) to get help?
27. How many girls (under age 18) do you know personally that have 
visited a Thuthuzela Care Centre (or were taken there) to get help?
28. How many boys (under age 18) do you know personally that have 
visited a Thuthuzela Care Centre (or were taken there) to get help?
29. Do you know of any care centers run by a non-governmental 
organization (NGO) in this community?
1=yes
2=no 
3=dk 
4=rf
A.2 There could be many reasons why a victim of sexual assault may not visit a Thuthuzela Care 
Center. We are interested in learning what some of these reasons are for people living in your 
community. 
For each of the following statements, please indicate how much you think each of the following is a 
barrier to visiting the Thuthuzela Care Centre, on a scale of 0-3 with “0” meaning not at all a barrier, 
“1” a minor barrier, “2” a barrier, and “3” meaning a major barrier. 
Responses: 0=Not a barrier, 1=A minor barrier, 2=A barrier, 3=A major barrier
30. Victims are not aware of the centers. 
31. Transportation challenges in going to the Thuthuzela 
Care Centre.
32. The Thuthuzela Care Centre is not open during 
convenient times. 
33. Feeling ashamed or embarrassed.
34. Feeling there is no one to trust at the Thuthuzela Care 
Centre.
35. Fear that the perpetrator would find out.
36. Fear that others in the community would find out.
37. Fear that people will blame the victim.
38. Fear that the victim will not receive the support she/he 
needs from the Thuthuzela Care Centre.
39. Fear that people at the Thuthuzela Care Centre will not 
believe the victim. 
40. Fear that the perpetrator will punish the victim.
41. The offer of money not to report the sexual assault.
42. What other reasons might prevent someone in your community from seeking 
assistance at the Thuthuzela Care Centre: 
_________________________________________
85
Now, please indicate how much you think each of the following is a barrier to reporting sexual 
offenses as crimes to the police, on a scale of 0-3 with “0” meaning not at all a barrier, “1” a minor 
barrier, “2” a barrier, and “3” meaning a major barrier. 
Responses: 0=Not a barrier, 1=A minor barrier, 2=A barrier, 3=A major barrier
43. Transportation challenges to the police station.
44. The police station is not open during convenient times. 
45. Feeling ashamed or embarrassed.
46. Feeling the police cannot be trusted.
47. Fear that the perpetrator would find out.
48. Fear that others in the community would find out.
49. Fear that people will blame the victim.
50. Fear that the victim will not receive the support she/he 
needs from the police.
51. Fear that the police will not believe the victim. 
52. Fear that the perpetrator will punish the victim.
53. Lack of information about where and to whom the 
incident should be reported. 
54. The offer of money not to report sexual violence.
55. What other reasons might prevent someone in your community from 
reporting a crime to the police: 
_________________________________________
B) In this community and elsewhere, people have different ideas about various issues in society, family, 
and relations between men and women. For the following list of statements, please state whether 
you strongly agree, agree, disagree or strongly disagree. 
Responses: 1=Strongly agree, 2=Agree, 3=Disagree, 4=Strongly disagree, 0=Don’t know
Statements Response
56. Rights for women mean that men lose out. 1=SA 2=A 3=D 
4=SD 0=DK 
57. There are times when a woman deserves to be beaten. 1=SA 2=A 3=D 
4=SD 0=DK 
58. A man should have the final word about decisions in his home. 1=SA 2=A 3=D 
4=SD 0=DK 
59. A woman should tolerate violence in order to keep her family 
together.
1=SA 2=A 3=D 
4=SD 0=DK 
60. A man and a woman should decide together what type of contraceptive 
to use.
1=SA 2=A 3=D 
4=SD 0=DK 
61. If a man sexually assaults his wife, others outside of the family should 
intervene.
1=SA 2=A 3=D 
4=SD 0=DK 
62. Would you agree or disagree that a man has the right to hit his wife in 
the following situations:
a) She does not complete her housework to his satisfaction
b) She disobeys him
c) She refuses to have sexual relations with him
d) He suspects that she is unfaithful
e) He finds out that she has been unfaithful
a) 1=SA 2=A 
3=D 4=SD 
0=DK 
b) 1=SA 2=A 
3=D 4=SD 
0=DK 
86
c) 1=SA 2=A 
3=D 4=SD 
0=DK 
d) 1=SA 2=A 
3=D 4=SD 
0=DK 
e) 1=SA 2=A 
3=D 4=SD 
0=DK 
63. Would you agree or disagree that a married woman can refuse to have 
sex with her husband in the following situations:
a) She doesn’t want to 
b) He is drunk
c) She is sick
d) He mistreats her
a) 1=SA 2=A 
3=D 4=SD 
0=DK 
b) 1=SA 2=A 
3=D 4=SD 
0=DK 
c) 1=SA 2=A 
3=D 4=SD 
0=DK 
d) 1=SA 2=A 
3=D 4=SD 
0=DK
64. When a woman is raped, she usually did something careless to put 
herself in that situation
1=SA 2=A 3=D 
4=SD 0=DK 
65. In some rape cases, women actually want it to happen 1=SA 2=A 3=D 
4=SD 0=DK 
66. If a woman doesn’t physically fight back, you can’t really say it was rape 1=SA 2=A 3=D 
4=SD 0=DK 
67. In any rape case, one would have to question whether the victim sleeps 
around a lot or has a bad reputation. 
1=SA 2=A 3=D 
4=SD 0=DK 
Sources: International Men and Gender Equality Survey, Women’s Survey (2008), section 2: attitudes about 
relations between men and women; WHO multi-country study on women's health and life events, version 10 
(2003); section 6: attitudes toward gender roles.
C) This section will ask you about your views regarding various issues related to experiences with sexual 
assault in your community. We are interested in your views regarding these statements. Please feel free 
to respond any way you like -- there are no right or wrong answers. 
Statements Response
68. Do you personally know any girls or women who have been raped or 
sexually assaulted in the last year? 
1=Yes
2=No 
3=Dk 
4=Rf
69. If yes, how many? _____
70. Do you know of any girls or women that have been raped or sexually 
assaulted in their lifetime?
1=Yes
2=No 
3=Dk 
4=Rf 
71. If yes, how many? _____
72. How often do women who are victims of sexual assault go to the 
police for assistance?
1=Always
2=Often 
3=Sometimes
87
4=Rarely
5=Never
73. Do you personally know any men or boys who have been raped or 
sexually assaulted in the last year?
1=Yes
2=No 
3=Dk 
4=Rf
74. If yes, how many? ____men (18+)
____boys (<18)
75. How often do men who experience sexual violence go to the police 
for assistance?
1=Always
2=Often 
3=Sometimes
4=Rarely
5=Never
76. How often do women who experience sexual violence seek medical 
treatment in a hospital, clinic, Thuthuzela Care Centere, or NGO 
crisis center?
1=Always
2=Often 
3=Sometimes
4=Rarely
5=Never
77. How often do men who experience sexual violence seek medical 
treatment in a hospital, clinic, Thuthuzela Care Centere, or NGO 
crisis center?
1=Always
2=Often
3=Sometimes
4=Rarely
5=Never
78. How often do learners (younger than 18) who experience sexual 
violence go to school teachers for assistance?
1=Always
2=Often
3=Sometimes
4=Rarely
5=Never
D) Case Scenarios
Scenario 1: A woman has to work late each night. The bus she takes home lets her off .5 km from her 
home. One night when walking home she is assaulted by a man. She is unable to fight him off and he rapes 
her. 
79. To what extent would you agree or disagree with the following 
statement: “Because of this woman’s behaviour, she was partially to 
blame for being sexually assaulted.”
1=Strongly agree 
2=Agree 
3=Disagree 
4=Strongly disagree
0=Don’t know
80. For women in this neighborhood who experience such situations, how 
likely is it that they would file a police report? 
1=Very likely
2=Likely
3=Not very likely
4=Not at all likely
81. Would the police be helpful in this situation? 1. Very helpful, 2. 
Helpful, 3. Not 
very helpful, 4. Not 
at all helpful
88
82. Among the following reasons, what would be the main reason that 
most women would not file a police report in this case? Please rank 
these 4 in order of importance.
a. Fear of being blamed
b. The police staff would not be helpful
c. Fear of the perpetrator
d. Lack of information about where to find help
____
83. Among the following reasons, what would be the second most 
important reason that most women would not file a police report in 
this case? Please rank these 4 in order of importance.
e. Fear of being blamed
f. The police staff would not be helpful
g. Fear of the perpetrator
84. Lack of information about where to find help
85. For women in this neighborhood who experience such situations, how 
likely are they to go to the Thuthuzela Care Centre? 
1=Very likely
2=Likely
3=Not very likely
4=Not at all likely
86. What would be the main reason that a woman would not go to the 
Thuthuzela Care Centre in this case? [Do not prompt] 
a. Fear of being blamed
b. The Centre staff would not be helpful
c. Fear of the perpetrator
d. Not aware of the TCC
e. Dk
____
87. Would the Thuthuzela Care Centre be helpful in this situation in any 
of the following ways? 
a. Medical 
b. Legal 
c. Psychological/emotional
a. 1. Very helpful, 2. 
Helpful, 3. Not 
very helpful, 4. Not 
at all helpful
b. 1. Very helpful, 
2. Helpful, 3. Not 
very helpful, 4. Not 
at all helpful
c. 1. Very helpful, 2. 
Helpful, 3. Not 
very helpful, 4. Not 
at all helpful
88. If you were in this situation, which of the following would you go to 
for help? 
a. Police: 
b. Hospital: 
c. Thuthuzela Care Centre: 
d. NGO crisis centre: 
e. Family member: 
f. Friend: 
g. Other: 
1=Definitely
2=Maybe
3=No 
4=dk 
5=rf
89
Scenario 2: An attractive 20 year old single woman wearing a mini-skirt goes out on a Friday night with 
friends. She stays for a few hours and has a few drinks. On her way home, she is assaulted by a man. She 
is unable to fight him off and he rapes her. 
89. To what extent would you agree or disagree with the following 
statement: “Because of this woman’s behaviour, she was to blame 
for being sexually assaulted.”
1=Strongly agree 
2=Agree 
3=Disagree 
4=Strongly 
disagree
0=Don’t know
90. For women in this neighbourhood who experience such situations, 
how likely is it that they would file a police report? =
1=Very likely
2=Likely
3=Not very likely
4=Not at all likely
91. Would the police be helpful in this situation? 1. Very helpful, 
2. Helpful, 
3. Not very 
helpful, 
4. Not at all 
helpful
a. Among the following reasons, what would be the main 
reason that most women would not file a police report in 
this case? Fear of being blamed
b. The police staff would not be helpful
c. Fear of the perpetrator
d. Lack of information about where to find help
e. None of the above: specify_________________
____
92. Among the following reasons, what would be the second most 
important reason that most women would not file a police report in 
this case? 
Fear of being blamed
The police staff would not be helpful
Fear of the perpetrator
Lack of information about where to find help
None of the above: specify_________________
93. For women in this neighbourhood who experience such situations, 
how likely are they to go to the Thuthuzela Care Centre? [Prompt]
1=Very likely
2=Likely
3=Not very likely
4=Not at all likely
94. What would be the main reason that a woman would not go to a 
Thuthuzela Care Centre in this case? [Do not prompt] 
a. Fear of being blamed
b. The Centre staff would not be helpful
c. Fear of the perpetrator
d. Lack of information about where to find help
e. dk
____
95. Would the Thuthuzela Care Centre be helpful in this situation in any 
of the following ways? 
a. 1. Very helpful,
2. Helpful, 
90
a. Medical 
b. Legal 
c. Psychological/emotional
3. Not very 
helpful, 
4. Not at all 
helpful 
b. 1. Very helpful, 
2. Helpful, 3. Not 
very helpful, 4. 
Not at all helpful
c. 1. Very helpful, 
2. Helpful, 3. Not 
very helpful, 4. 
Not at all helpful
96. If you were in this situation, which of the following would you go to 
for help? 
h. Police: 
i. Hospital: 
j. Thuthuzela Care Centre: 
k. NGO crisis centre: 
l. Family member: 
m. Friend: 
97. Other: 
1=Definitely
2=Maybe
3=No 
4=dk 
5=rf
G) Now I am going to ask a few questions about you.
98. What is your age?
99. What is your race? 1=Black 2=White 
3=Coloured 4=Indian 
5=Asian 
6=other: 
___________________ 
100. What is the primary language spoken at your home? 
(Do not, prompt]
1. Afrikaans
2. English
3. Ndebele
4. Pedi
5. Sotho
6. Swati
7. Tshonga 
8. Tswana
9. Venda
10. Xhosa
11. Zulu
12. Other: 
___________________
101. What is the highest level of education you have 
completed? 
(Do not, prompt]
1=No formal schooling
2=Preschool
3=Primary school
4=Secondary School
5=Tertiary/FET
91
6=Graduate or higher
102. How much money did your household earn last 
month? (R per year)
1=Less than 1,000
2=1,000-5,000 
3=5,001- 10,000 
4=More than 10,000
92
Supplemental TCC Intake Form
IMPACT EVALUATION INTAKE FORM
PART 1 (MANDATORY FOR 1ST VISIT) 
*TO BE COMPLETED BY TCC SITE MANAGER FOR ALL SURVIVORS PRESENTING TO TCC STARTING IN AUGUST 
2014
1. TCC Name: 
_________________________________________________________________ 
2. Date survivor presented to TCC: _______/________/________Time (00:00h):_______:______
(DD) (MM) (YYYY) (hh) (mm)
Question Response 
3. Is this the survivor’s first visit to 
this TCC? Yes___ No___ Unknown___ Refused response ___
4. If no, when was the survivor’s 
first visit to this TCC?
_________/_________/_______________ N/A (1st visit) 
(DD) (MM) (YYYY)
5. Type of crime being reported:
Rape__ Attempted rape__ Other sexual assault __ Domestic 
Violence Other __ Unknown/not reported __
6. Incident date and time (this 
incident):
_________/_________/_______ Time: _____:__________
(DD) (MM) (YYYY) (hh) (mm)
7. Was this survivor brought to 
the TCC by the police? Yes___ No___ Unknown___ Refused response ___
8. If yes: Police station where crime 
occurred:
9. If yes: Police station where 
survivor resides: 
10. If no: Address/neighborhood 
where the crime occurred: 
Police station closest to that 
address/neighborhood: 
11. If no: Address/neighborhood 
where survivor resides: 
Police station closest to that 
address/neighborhood: 
12. Sex of survivor: male___ female___ 
13. Age of survivor:
0-5 yrs__ 6-11 yrs__ 12-17 yrs__ 18-27 yrs__ 26-35 yrs__ 
36-45 yrs__ 46-55 years__ 56 yrs or older__ dk__
14. Date of Birth of survivor:
_________/_________/________ Unknown/refused ___
(DD) (MM) (YYYY)
15. If no, who physically brought the 
survivor to the TCC?
N/A (brought by police) ___ Healthcare 
professional___Teacher___ TCC Staff___ Legal 
93
Professional___ NGO worker___ Family member/friend__
Nobody (survivor came alone) ____
Other ____ Specify: 
_______________________________________
16. Regardless of how the survivor 
arrived at the TCC, was the 
survivor referred or 
recommended to the TCC by 
any of the following?
 [Mark all that apply]
Police___ Healthcare professional___ Teacher___ TCC 
Staff___ 
Legal Professional___ NGO worker___ Family or friends ____ 
Other___ None____
Specify (NGO/Other): 
______________________________________
Services Planned? Completed? Results Received?
17. Forensic exam Yes__ No__ Yes__ No__ Yes__ No__ 
18. Consultation Yes__ No__ Yes__ No__ Yes__ No__ 
19. Post-exposure prophylaxis Yes__ No__ Yes__ No__ Yes__ No__ 
20. Litigation Yes__ No__ Yes__ No__ Yes__ No__ 
21. Safe place to stay Yes__ No__ Yes__ No__ Yes__ No__ 
22. Referrals made? If yes, to: 
________________________
__________________ Yes__ No__ Yes__ No__ Yes__ No__ 
23. Other, please specify: 
________________________
__________________ Yes__ No__ Yes__ No__ Yes__ No__ 
IMPACT EVALUATION INTAKE FORM
PART 2 (2ND VISIT/2ND COUNSELING SESSION) 
* TO BE COMPLETED WITH THE SURVIVOR (OR PARENT/GUARDIAN/FAMILY MEMBER) IF HE/SHE IS WILLING.
MUST BE ADMINISTERED IN PRIVATE.
1. The respondent is: The survivor___ Parent/guardian___ Other family member___ Other___
2. Date survivor first presented to TCC: _______/________/___________ 
 (DD) (MM) (YYYY) 
3. Survivor Date of Birth: _______/________/___________ 
 (DD) (MM) (YYYY) 
Please ask the survivor (or parent/guardian/family member) the following questions.
Question Response 
4. In the past three months, have you 
heard or seen any advertisements, 
announcements, or spots promoting 
awareness of resources available to 
survivors of sexual violence?
[If no, skip to question 6] Yes___ No___ Unknown___ Refused response ___
5. If yes, what were the main messages 
do your remember?
[Do not prompt, mark all that 
apply.]
Awareness of sexual assault as a problem___
Where to seek help___ Rights of victims___
Messages to deter perpetrators___ Other___
Unknown___ Refused response___
94
Specify 
(Other):______________________________________
6. What are the main difficulties that 
you and people you know encounter 
in accessing TCCs? 
7. Has the TCC been open during the 
times you have wanted to come to 
the TCC? Yes___ No___ Unknown___ Refused response ___
8. How did you get to the TCC 
center?
Public transport___ Private car___ Walked___
Driven by police___ Other___
Specify 
(Other):______________________________________
9. Have you told teachers or any adult 
at school about this incident (if in 
school)?
Yes___ No___ Unknown___ Refused response ___ 
N/A ___
10. Have you told police about this 
incident?
Yes___ No___ Unknown___ Refused response ___ 
11. How would you rate the level of 
respect shown to you by the police?
[Go to end]
Disrespectful___ 
Somewhat disrespectful___
Respectful___ 
Very respectful___
Refused response___
12. Why did you not go to the police? 
[Do not prompt. Mark all that apply]
Transportation challenges to the police station___
Fear that the police will not believe you___
Feeling the police cannot be trusted___
Feeling that police reporting is not useful___
Feeling ashamed or embarrassed___
Fear that others in the community would find out___
Fear that the attacker would find out___
Fear that the attacker will punish you___
Fear that you will not receive the support you need from the police___
The TCC is more helpful___
The offer of money not to report this to the police___
Other___
Specify (other):_________________________________________________
95
NGO Interview Guide
(Please remember, this is just a guide. Please be sure to follow-up on interesting points with additional 
questions.) 
• Basic information about the NGO: 
o Please tell me about what your organization does with respect to survivors of sexual 
assault and gender-based violence? 
o Phone number and days/hours of operation 
o What is your organization’s relationship with the TCC? 
• TCC capacity: 
o What is your general impression of the capacity of the _____ TCC? 
o Do you think the TCC is understaffed or under-resourced in any way? 
o Is the TCC able to meet the needs of the survivors who present there? 
o Is the TCC able to meet the needs of this community with regard to sexual assault and 
gender-based violence? 
o What challenges do you think the TCC experiences in serving survivors? 
• Current TCC use: 
o What do survivors who need help after hours working hours usually do? (e.g., go the 
hospital, wait in police station, etc.) On a weeknight? On Friday? On weekend? 
o How do survivors get to the TCC – what are the transport options? Which proportion 
of survivors use each method of transport? Do you think survivors view transportation 
as a barrier? How do they pay for it? Does the TCC reimburse transport expenses for 
survivors? 
o What is the typical time a survivor waits to be helped by TCC staff? And the longest 
time? For those with long wait periods, what is the reason? 
o How well known is the TCC in the surrounding communities? 
o Do people understand the services offered there? 
o Are there any common misperceptions about the TCC that you know of? 
o Why do you think some survivors do not go to the TCCs? 
o Do you have a sense of how frequently survivors go to the TCC because of a 
professional referral? What kinds of professionals are these? Which are the most 
common? 
o Are survivors sometimes taken to the TCC by other organizations/non-profits? If so, 
which NGOs help survivors who come to your TCC? 
o How often do female survivors over the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of female/adult survivors) 
o How often do female survivors under the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of female/child survivors) 
o How often do male survivors over the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of male/adult survivors) 
o How often do male survivors under the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of male/child survivors) 
o How often do lesbian/gay/bisexual/transgender (LGBT) survivors come to the TCC for 
assistance? (Typical number of cases per week/month; percentage of LGBT survivors) 
o Do you think the majority of all survivors feel comfortable going to TCC? 
o Do some types of survivors experience higher barriers/challenges to seeking help than 
others? Which groups experience the most barriers? Why do you think that is? 
96
o What particular features of the community have the largest impact on sexual assault and 
gender-based violence (GBV)? Are there certain aspects of the cultural context in 
community that you can describe that affect GBV prevalence? 
o Are there certain survivor needs that the TCC is not able to address? What are some 
examples of these needs? 
97
TCC Interview Guide
Protocol for TCC Baseline Visit: 
• Prepare for TCC visit 
o A data collection firm researcher will call the TCC in advance of fieldwork to set an 
appointment for a visit, which is expected to take approximately half a day. The purpose 
of the visit will be briefly described and availability of data to be obtained during the visit 
should be confirmed. Contact information for the staff at each of the TCCs will be 
provided by SI. A letter of support from NPA will also be sent to the TCC contact 
person to confirm approval for the visit. The visit should be reconfirmed the day prior 
and rescheduled, if necessary. 
• Visit TCC 
o A flash drive and secure folders should be brought to the visit for downloading data 
and/or transporting hard-copy data. 
o When the researchers arrive at the TCC, they will contact the primary point-of-contact 
to explain the purpose of the visit and show documentation of NPA’s support for the 
visit (i.e., support letter). At this time, the researchers should also explain the agenda 
for the meeting: 
 Collect basic information about the TCC. (See questions on tablet.) 
 Explain the Supplemental Intake Form. 
o Data: Ask the TCC staff if any electronic records are available. If so, the data download 
process should be started at the beginning of the visit to ensure that the data copying 
can be completed before the visit is over. Names of survivors should be removed prior 
to download. If electronic data is not available, then hard copies of the files should be 
copied or scanned.
o TCC information: Interview the TCC staff to collect relevant information about the 
TCC, posing the questions listed below. This interview should be audio recorded. 
o Supplemental Intake Form: While the data is being downloaded, invite all TCC staff 
present and available to join a short training on the new data collection tool. Explain the 
Supplemental Intake Form and answer any questions that may arise. Ensure that 
additional questions are added to existing Intake Forms and that columns are added in 
existing electronic databases to accommodate the new information (if the TCC enters 
data into the computer). Both hard and soft copies of the new questions should be 
provided.
Interview Guide: 
• Basic TCC information: 
o Phone number and days/hours of operation 
o Staff positions at the TCC, which ones are currently filled and when they were filled (by 
anyone, not necessarily when the existing staff joined). 
o How long has the TCC been adequately staffed? 
o TCC resources (i.e., computer, internet, phone, office supplies, medical supplies, 
examination office, consultation space, waiting room, information for survivors, etc.) 
o Notes: 
• TCC capacity: 
o Are there currently plans to add any staff to this TCC? If so, what positions. 
o Are there enough staff at this TCC to serve survivors who come here? 
o Which staff would you like to see added, if any? (List positions) 
98
o If you know, can you please tell me who provides the funding for the staff of TCC? (e.g., 
donors, government) 
o Does this TCC have sufficient resources to meet the needs of survivors who come 
here? [Select one: Never, Rarely, Sometimes, Often, Always] 
o What are the main challenges this TCC experiences in serving survivors? 
o General notes on capacity:
• Current use: 
o What do survivors who need help after hours working hours usually do? (e.g., go the 
hospital, wait in police station, etc.) On a weeknight? On Friday? On weekend? 
o Is a survivor required to file a police report before receiving service from the TCC? 
What happens if a survivor comes to the TCC without filing a police report? Is s/he 
taken or asked to go to the police station or can s/he be treated first? Can a survivor 
refuse to file a police report and still get service from the TCC? 
o Here we have a list of police stations and satellite stations serve your TCC. Are there 
any others? Are some of the police stations more active in helping survivors? Can you 
discuss the differences? 
o How do survivors get to the TCC – what are the transport options? Which proportion 
of survivors use each method of transport? Do you think survivors view transportation 
as a barrier? How do they pay for it? Does the TCC reimburse transport expenses for 
survivors? 
o What is the typical time a survivor waits to be helped by TCC staff? And the longest 
time? For those with long wait periods, what is the reason? 
o How well known is the TCC known in the surrounding communities? 
o Do people understand the services offered here? 
o Are there any common misperceptions about the TCC that you know of? 
o Why do you think some survivors do not come to the TCCs? 
o Do you have a sense of how frequently survivors come to the TCC because of a 
professional referral? What kinds of professionals are these? Which are the most 
common? 
o Are survivors sometimes taken to the TCC by other organizations/non-profits? If so, 
which NGOs help survivors who come to your TCC? 
o How often do female survivors over the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of female/adult survivors) 
o How often do female survivors under the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of female/child survivors) 
o How often do male survivors over the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of male/adult survivors) 
o How often do male survivors under the age of 18 come to the TCC for assistance? 
(Typical number of cases per week/month; percentage of male/child survivors) 
o How often do lesbian/gay/bisexual/transgender (LGBT) survivors come to the TCC for 
assistance? (Typical number of cases per week/month; percentage of LGBT survivors) 
o Do you think the majority of all survivors feel comfortable coming to TCC, or do some 
types of survivors experience higher barriers to seeking help than others? Which groups 
experience the most barriers? Why do you think that is? 
o What are particular features of the community that have the largest impact on gender￾based violence (GBV) in this area? Are there certain aspects of the cultural context in 
community that you can describe that affect GBV prevalence? 
o Are there certain survivor needs that the TCC is not able to address? What are some 
examples of these needs? 
99
ANNEX III: WORKS CITED
Abrahams, Naeemah, Rachel Jewkes, Lorna J. Martin, Shanaaz Mathews, Lisa Vetten, and Carl Lombard. 
2009. “Mortality of Women from Intimate Partner Violence in South Africa: A National 
Epidemiological Study.” Violence and Victims. 24(4): 546–56. 
Anderson, Linda A., and Susan C. Whiston. 2005. “Sexual Assault Education Programs: A Meta-Analytic 
Examination of Their Effectiveness.” Psychology of Women Quarterly. 29: 374–388. 
Angrist, Joshua D., and Jörn-Steffen Pischke. 2010. “The Credibility Revolution in Empirical Economics: 
How Better Research Design is Taking the Con out of Econometrics.” Journal of Economic 
Perspectives. 24(2): 3–30. 
Banerjee, Abijit, and Esther Duflo. 2008. “The Experimental Approach to Development Economics.” 
Working Paper, National Bureau of Economic Research.
Duflo, Esther, Rachel Glennerster, and Michael Kremer. 2006. “Using Randomization in Development 
Economics Research: A Toolkit.” Centre for Economic Policy Research, Working Paper.
Gibbs, Andrew, Patience Mungwari Mpani, Deborah Ewing, Leandri Pretorius, Samantha Willan, and Kay 
Govender. 2014. “Understanding Barriers and Opportunities to the Use of Medico-Legal Evidence 
in the Prosecution of Sexual Violence in South Africa.” Draft Research Proposal.
Gender Links and Medical Research Council. 2011. “The War @ Home. Preliminary Findings of the 
Gauteng Gender Violence Prevalence Study.” 
Imai, Kosuke, Gary King, and Elizabeth A. Stuart. 2007. “Misunderstandings between Experimentalists and 
Observationalists about Causal Inference.” Technical Report.
Jewkes, Rachel, and Naeemah Abrahams. 2002. “The Epidemiology of Rape and Sexual Coercion in South 
Africa: An Overview.” Social Science and Medicine. 55: 153–166. 
Jewkes, Rachel, and Robert Morrell. 2010. “Gender and Sexuality: Emerging Perspectives from the 
Heterosexual Epidemic in South Africa and Implications for HIV Risk and Prevention.” Journal of 
the International AIDS Society. 13(6): 1-11. 
Jewkes, Rachel, Yandisa Sikweyiya, Robert Morrell, and Kristin Dunkle. 2011. “Gender Inequitable 
Masculinity and Sexual Entitlement in Rape Perpetration South Africa: Findings of a Cross￾Sectional Study.” PLoS ONE. 6(12): e29590. 
Kasy, Maximilian. 2013. “Why experimenters should not randomize and what they should do instead.” 
Unpublished working paper.
Shai, Nwabisa Jama, and Yandisa Sikweyiya. 2015. “Programmes for Change: Addressing Sexual and 
Intimate Partner Violence in South Africa.” SA Crime Quarterly. 51: 31–41. 
Soul City Institute. 2013. “Qualitative formative research on the knowledge, attitudes, and behaviors 
relating to reporting sexual assault and the use of Thuthuzela Care Centers.”
100
South African Police Service. 2014. “Crime Statistics: April 2013 - March 2014.” 
http://www.saps.gov.za/resource_centre/publications/statistics/crimestats/2014/crime_stats.php. 
101
ANNEX IV: WOMEN’S SURVEY SAMPLING PROTOCOL 
Sampling sectors and subsectors
The Stakeholder Engagement Officer, a former member of SAPS will access precinct maps from police 
stations to establish: 
• The sectors and subsectors within the precincts. 
• Insights from maps and police officials which will enable the determination of: 
• Areas in the precincts that are primarily industrial. These areas will be excluded. If a sector 
(or subsector) is partially industrial and partially residential, the entire sector (or 
subsector) is not excluded. Rather, the industrial area within that sector/sub-sector 
should excluded. 
• Areas in the precincts that are inhabited by affluent residents who would have private 
insurance and therefore not use a TCC (mostly former white-only suburbs, with higher 
income residents). These areas will also be excluded. If a sector (or subsector) is partially 
high-income and partially low-income, the entire sector (or subsector) was not excluded. 
Rather, the high-income area within that sector/sub-sector is to be excluded. 
• The sample should be derived from the remaining sectors and subsectors that were not excluded. 
Sampling individual households
Procedures followed for Individual Households: 
• In cases where the number of sub-sectors were less than 15, one household should be selected 
from each of the available sub-sectors of a precinct and an additional 4 households selected from 
4 randomly selected precincts from the same 11 precincts. In cases where the sub-sectors per 
precinct are more than 15, 15 sub-sectors are to be randomly selected and one household 
selected from each sub-sector following the procedure outlined above. These 15 sub-sectors 
should be evenly distributed between sectors (e.g. in cases where there are three sectors, five 
sub-sectors are drawn from each). One exception to this rule is when the urban/rural weighting 
rules suggested otherwise. For example, if three urban households and 12 rural households are 
selected and one sector is distinctly urban and the other two are distinctly rural, according to 
weighting data only three households should be selected from the urban sector and 12 from the 
rural sector. 
• Sampling procedures for each precinct should always discussed and approved by the team BEFORE 
sampling is done by the Fieldwork Coordinator, and cross-checked for adequacy before the 
fieldworkers were deployed. This requires constant communication between the Fieldwork 
Coordinator and the Fieldwork Manager, with oversight guidance being provided by the Project 
Leader. 
• Before data collection, the sampled HH coordinates are submitted and approved by SI. 
In-house sampling
• In households where there were two or more eligible females, a raffle is conducted to randomly 
select a participant. 
• Every female (aged 18 to 49) who resided in the selected household during the time of the visit 
must be included in the raffle. If the selected person is not available, the team should come back 
at a later stage when they are available. A total of three attempts are made to reach the sampled 
individual before resampling a secondary individual. 
 
102
ANNEX V: COMPLETENESS OF BASELINE SERVICE PROVIDER DATA
Variable
n (observations with non￾missing data)
# of missing 
responses
% of responses 
missing
A3 1,737 52 3%
A4 1,753 36 2%
A5 1,735 54 3%
A7 1,753 36 2%
A8 1,754 35 2%
A10 1,756 33 2%
A11 1,762 27 2%
A12 1,737 52 3%
A13 1,695 94 5%
A14 1,686 103 6%
A15 1,676 113 6%
A16 1,710 79 4%
A17 1,707 82 5%
B1 1,682 107 6%
B2 1,663 126 7%
B3 1,638 151 8%
B4 1,657 132 7%
B5 1,664 125 7%
B6 1,657 132 7%
B7 1,612 177 10%
B8 1,644 145 8%
B9 1,663 126 7%
B10 1,673 116 6%
B12 1,619 170 10%
B14 1,699 90 5%
C1 1,739 50 3%
C2 1,732 57 3%
C3 1,721 68 4%
C4 1,720 69 4%
C5 1,724 65 4%
D1 1,640 136 8%
D2 1,736 49 3%
103
D3 1,710 70 4%
D4 1,695 86 5%
D5 1,687 87 5%
D6 1,708 73 4%
D7 1,682 95 5%
D8 1,684 96 5%
D9 1,675 99 6%
E1 1,706 75 4%
E2 1,684 95 5%
E3 1,687 90 5%
E4 1,687 91 5%
E5 1,670 107 6%
E6 1,628 150 8%
E7 1,649 131 7%
E8 1,646 134 7%
E9 1,652 123 7%
E10 1,101 688 38%
F1 1,717 72 4%
F2 1,721 68 4%
F3 1,706 83 5%
104
ANNEX VI: REGRESSION TABLES
Logistic regression results table for effect of Treatment 1 on knowledge of TCCs (n=2,950)
Have you heard of the TCC?
Odds 
ratio p-value
Std. 
error
DiD Treatment effect 1.036 0.903 0.302
Treatment 0.960 0.840 0.195
Endline 1.118 0.606 0.243
Age 1.019*** 0.008 0.007
Zulu 1.019 0.941 0.253
Xhosa 0.620* 0.097 0.179
Tswana 0.922 0.743 0.229
Afrikaans 1.184 0.666 0.463
Coloured 0.663 0.227 0.226
Lower than secondary education 0.546*** 0.000 0.080
Tertiary education or higher 1.219 0.217 0.196
Income < R1,000 per month 1.693*** 0.000 0.255
Income > R5,001 per month 1.208 0.199 0.178
Urban precinct 0.840 0.245 0.126
Knows a SGBV survivor 4.811*** 0.000 0.581
Perception of SGBV as a problem in the community 1.123* 0.089 0.077
Sex crimes per 1,000 1.293** 0.017 0.139
Woman's right to refuse sex PCA 0.973 0.348 0.028
Tolerance of violence toward women PCA 1.072** 0.023 0.033
Rape victim-blaming PCA 1.052 0.208 0.043
province: Limpopo 7.733*** 0.000 4.228
province: North We 5.288*** 0.001 2.582
province: Kwa-Zulu 4.499*** 0.001 2.098
province: Eastern 17.06*** 0.000 8.442
province: Western 5.946*** 0.001 3.230
province: Nothern Cape 6.547*** 0.000 3.051
province: Free State 7.209*** 0.000 3.546
province: Mpumalanga 2.661* 0.076 1.466
Constant 0.0063*** 0.000 0.003
Observations 2950
Pseudo R2 0.176 
Robust p-value in parentheses
*** p<0.01, ** p<0.05, * p<0.1
105
Regression results table for impact of community dialogues on attitudes toward SGBV (n=2,931)
Attitudes Index (PCA) Coefficient
p-value
Std. 
error
DiD Treatment effect 0.276 0.285 0.257
Treatment -0.285* 0.071 0.157
Endline 0.649*** 0.000 0.180
Age 0.0149*** 0.008 0.006
Zulu 0.526*** 0.003 0.171
Xhosa 0.166 0.515 0.254
Tswana -0.091 0.722 0.255
Afrikaans 0.564 0.284 0.524
race: white -0.379 0.384 0.434
race: coloured -0.296 0.525 0.464
race: indian 0.381 0.228 0.314
race: asian 3.615*** 0.000 0.257
Lower than secondary education -0.864*** 0.000 0.144
Tertiary education or higher -0.824*** 0.000 0.173
Income < R1,000 per month -0.236 0.137 0.158
Income > R5,001 per month 1.107*** 0.000 0.148
Sex crimes per 1,000 -0.11 0.245 0.094
Perception of SGBV as a problem in the community -0.176** 0.014 0.070
Knows a SGBV suvivor 0.577*** 0.000 0.106
province: Gauteng 0.0293 0.917 0.280
province: Limpopo -0.079 0.788 0.292
province: North West 0.435 0.242 0.370
province: Kwa-Zulu 0.145 0.652 0.320
province: Western Cape -0.498** 0.026 0.221
province: Nothern Cape -0.335 0.264 0.298
province: Free State 0.343 0.349 0.365
province: Mpumalanga 0.349 0.264 0.311
Urban precinct 0.137 0.288 0.128
Constant -0.779** 0.031 0.356
Observations 2,931
R-squared 0.118
Robust standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
106
ANNEX VII: INITIAL POWER CALCUATIONS
Power Calculations 
One of the challenges of this evaluation is the small sample size of geographic units available for randomized 
sampling. The number of TCC catchment areas is relatively small since there are only 51 TCC currently 
in operation. For the purposes of the evaluation, a community is defined as a police precinct, following 
extensive consultation with government officials and CSO’s and the determination that this is a natural 
community unit at which local administrative data is available. Many TCCs have only three to six police 
stations serving them, so for each TCC, three of the neighboring police precincts will be randomly sampled 
into the study. Due to these small numbers, the design had to be limited to only two interventions with 
one control. While it would have been useful to evaluate the impact of other interventions, or have a 
third treatment group with the two interventions implemented at the same time to learn more about the 
combined effects, the small sample size of TCCs and communities within each TCC catchment area made 
these options infeasible.
For the purposes of the power calculation, each group is rounded to 50, so the modeled design calculates 
MDES assuming ~50 sites per group. 
The first project component is a two-level multi-site cluster randomized trial (MSCRT) with cluster level 
outcomes. In this case, randomization occurs at the second level, which is the community. The outcomes 
for this component are also measured at the community level.
The second project component is three-level MSCRT with individual-level outcomes. Here, again, 
randomization occurs at the second level, the community, but in this case, the outcomes are measured at 
the first level: the level of the individual. 
For both designs, blocking shall be used in order to improve the precision and power of the IE. Depending 
on data availability, several of the following types of variables will be used to construct site blocks: 
geographic regions, rural versus urban, crime rates and reported incidents of sexual assault, and 
demographics. These sites or blocks can be viewed as fixed. 
Unfortunately, since few comparable studies are available and there is no possibility to conduct a pilot, it 
is difficult to have a high degree of confidence in the estimates of model parameters and the corresponding 
effect sizes. The following calculations use relatively conservative estimates for each parameter for this 
reason.
All power calculations were performed using the Optimal Design software package. 
Design 1: Cluster-Level Outcomes
The first design is a two-level MSCRT, or blocked, cluster randomized trial. The first level represents the 
stage in which communities shall be blocked prior to randomization, and the second level represents the 
stage where the interventions will be administered to the treatment communities, after which cluster level 
outcomes will be measured. These outcomes include precinct-level estimates of the total number of GBV 
reports, utilization of TCC services, etc. obtained from administrative data. The power calculations are 
based on standard assumptions, which are presented in Table 1.
107
Table 1: Assumptions for Power Calculations, Design 1
Target power: 80%
α = 5%
Cluster Level Reliability (CLR) = .70 (Publishable standard)
В = .1, .2, .3 (*To be adjusted when data is received)
σ2 = 0 (fixed effects)
К = 2
J = 50
δ = standardized MDES, to be estimated
This IE employs an effect size approach to power analysis because the number of communities that can 
be used in the intervention is immutable. First, the desired level of power is set to .80 and significance 
level to .05 (as is standard in the literature). Next, the effect size that can be detected for the given sample 
size can be computed. There are fifty communities in each group, and we use two blocking sites to ensure 
an even number of clusters across each site. No adjustment is made for inclusion of any covariates for the 
purposes of these calculations. If covariates are included, this will serve to slightly improve the power. 
The calculations use the default .70 cluster level reliability (since this is the standard for publication). 
Table 2 presents the results for MDES estimates under the assumptions listed in Table 1. A graph showing 
the MDES versus power is presented in Figure 4. The MDES ranges from 0.53 and 0.65, which is 
comparable to a moderate treatment effect.
Table 2: Summary of MDES under Various Assumptions for Design 1
Power Alpha CLR B MDES
.80 .05 .7 .1 .65
.80 .05 .7 .2 .63
.80 .05 .7 .3 .60
.80 .10 .7 .1 .58
.80 .10 .7 .2 .56
.80 .10 .7 .3 .53
108
Figure 1: Relationship between MDES and Power, Design 1, cluster level outcomes
Design 2: Individual-Level Outcomes
The second design is a three-level MSCRT, or blocked cluster randomized trial with individual level 
outcomes. Specifically, the first level is the blocking variable, the second level is the treatment level 
(clusters - communities), and the third level is the measurement level (individuals). The outcomes will be 
gathered using surveys to be conducted in each community. The surveys will poll twenty female household 
heads per community and twenty institutional representatives. Table 3 shows the standard assumptions 
utilized by the power calculations. 
Again baseline data or pilot results are not available, so we opted to use .1 and .2 for the ICC coefficients 
among individual, more conservative than Optimal Design’s default of .05. 
The power calculations for the survey instruments do not assume a panel survey or repeated measures, 
which would increase the power of the study since data collection will not be panel-based. Instead, new 
respondents will be drawn each round. 
109
Table 3: Assumptions for Power Calculations, Design 2
Target power: 80%
α = 5%
Intra-Class Correlation = .1 to .2
В = .2 and .3
σ2 = 0 (fixed effects)
К = 2
J = 50
η1 = 20 surveys (institutional) 
η2 = 20 surveys (community - household) 
δ = standardized MDES, to be estimated
Table 4 displays the estimated MDES under a range of different assumptions, with variation on the amount 
of variation explained by the blocking variable “B” from 20-30 percent, and ICC among households from 
.1 to .2. These calculations yield an MDES for individual level outcomes of between .19 and .26, significantly 
lower than for the first design. This design is able to capture small effects. Figure 5 demonstrates the 
relationship between MDES and power for the second design.
Figure 2: Relationship between MDES and Power – Design 2, individual level outcomes
110
Table 4: Summary of MDES under Various Assumptions for Design 2
Assumptions of the Sampling and Randomization Plan
The feasibility of the implementation of the proposed sampling and randomization plan is dependent on 
the following two assumptions: 
1. The implementing partners will be able to complete the implementation within the boundaries of 
the one police precinct in each TCC catchment area assigned to the relevant treatment group.
2. The evaluation team will receive, in a timely manner, TCC case record data that is or can be 
disaggregated to police precinct level. 
Should one of these assumptions not hold, the evaluation team would need to implement an alternative 
plan of randomizing at the TCC level. It must be noted that TCC-level randomization would result in a 
loss of power for the IE. The loss of power would be most significant for design 1 (precinct-level 
outcomes). The power calculations for TCC-level randomization yield a much higher range for the MDES 
(.86 to .93, depending on the assumptions), implying only large treatment effects of the intervention would 
be able to yield statistically significant measured differences within this IE design. For design 2 (individual￾level outcomes), the change in MDES is less dramatic, though still notable (.27 to .37 under varying 
assumptions). These options are presented in further detail in table 5 below.
Table 5: Overview of possible approaches
Preferred option Alternative option
Problem and 
research 
question 
TCC utilization is lower than desirable given the extent of gender based violence. 
How then should utilization be increased? More specifically, what are the 
independent effects of the following two approaches to increasing TCC utilization: 
(1) an educational campaign targeted at women in TCC catchment areas designed to 
inform them about the TCCs and reduce the stigma attached to using TCCs, and (2) 
a training program designed for police and teachers to help them identify victims of 
gender based violence and refer them to the TCCs.
Approaches Within each of the TCC catchment 
areas SI will randomly assign police 
precincts to one of three groups: 
control, professional referrals, or 
educational campaign groups. (Typically 
there are 3-6 precincts per catchment 
area.) As such, the evaluation will be 
based on 153 police precincts, including 
51 control, 51 professional referrals, 
and 51 education campaign precincts. 
The main evaluation will compare TCC 
utilization at the precinct level before 
and after the intervention and 
determine which approach, if either, 
SI will randomly assign the 51 TCCs to 
one of three groups: 17 will be assigned to 
the control group, 17 to the professional 
referral group, and 17 to the educational 
campaign group. Given the size of the 
catchment areas (often a radius of 25km 
around the TCC) it will still be desirable 
to focus the intervention in specific 
precincts. The main evaluation will 
compare TCC utilization at the precinct 
level before and after the intervention and 
determine which approach, if either, 
resulted in greater increases in TCC 
utilization.
Power Alpha N B ICC MDES
.80 .05 20 .20 .20 .26
.80 .05 20 .30 .20 .24
.80 .05 20 .20 .10 .20
.80 .05 20 .30 .10 .19
111
resulted in greater increases in TCC 
utilization. 
Strengths/
weaknesses
This approach will provide a higher 
number of observations to test the 
impact of the two approaches. The 
design will be able to detect a 
moderate increase in TCC utilization. 
Nonetheless this approach is only 
possible if the following conditions are 
met. (1) TCCs must collect and 
provide intake information at the 
precinct level, (2) the implementing 
partners must be able to conduct 
activities in all 51 TCC catchment areas 
and limit their activities to assigned 
precincts. 
This approach will be easier and more
cost effective for implementation of the 
intervention. Nonetheless, because of the 
small sample size, the main impact 
evaluation will be poorly powered, and 
will only be able to detect a large increase 
in TCC utilization. This will be a high bar 
to overcome and it is unlikely that the IE 
will be able to make a confident 
determination. Even under this approach, 
it will still be desirable for the TCCs to 
collect and provide intake information at 
the precinct level. The study will still be 
able to confidently test aspects of the 
education campaign through the 
household survey.
112
ANNEX VIII: COMMUNITY SURVEY DATA COLLECTION TIMELINE
IRI HOUSEHOLD DATA COLLECTION
Province TCC Police Precinct Baseline Date
Endline 
Date
Gauteng Laudium Diepsloot 14-Aug-14 3-Feb-15 
Gauteng Laudium Olievenhoutbosch 15-Aug-14 3-Feb-15
Gauteng Tembisa Rabie Ridge 19-Aug-15 4-Feb-15
Gauteng Tembisa Thembisa 20-Aug-15 4-Feb-15
Gauteng Natalspruit Thokoza 21-Aug-14 9-Feb-15
Gauteng Natalspruit Vosloorus 22-Aug-14 9-Feb-15
Gauteng Mamelodi Mamelodi 12-Aug-14 2-Feb-15
Gauteng Mamelodi Silverton 23-Aug-14 2-Feb-15
Gauteng Baragwanath Meadowlands 25-Aug-14 10-Feb-15
Gauteng Baragwanath Orlando 26-Aug-14 10-Feb-15
Gauteng Lenasia Lenasia 27-Aug-15 6-Feb-16
Gauteng Lenasia Ennerdale 27-Aug-14 6-Feb-15
Gauteng Kopanong Vanderbljipark 28-Aug-14 5-Feb-15
Gauteng Kopanong Sebokeng 28-Aug-14 5-Feb-15
Limpopo Mankweng Mankweng 6-Sep-14 1-Mar-15
Limpopo Mokopane Mahwelereng 3-Sep-14 26-Feb-15
Limpopo Mokopane Tinmyne 3-Sep-14 27-Feb-15
Limpopo Seshego Mashashane 5-Sep-14 28-Feb-15
Limpopo Seshego Seshego 5-Sep-14 1-Mar-15
Limpopo Nkhensane Giyani 9-Sep-14 2-Mar-15
Limpopo Musina Musina 10-Sep-14 3-Mar-15
Limpopo Musina Masisi 11-Sep-14 4-Mar-15
Limpopo Thohoyandou Thohoyandou 12-Sep-14 3-Mar-15
Limpopo Thohoyandou Magatle 13-Sep-14 28-Feb-15
Mpumalanga Tonga Komatipoort 19-Nov-14 10-Mar-15
Mpumalanga Tonga Tonga 20-Nov-14 10-Mar-15
Mpumalanga Temba Pienaar 21-Nov-14 11-Mar-15
Mpumalanga Temba Kabokweni 22-Nov-15 11-Mar-15
Mpumalanga Ermelo Breyton 23-Nov-15 12-Mar-15
Mpumalanga Ermelo Ermelo 25-Nov-14 12-Mar-15
Mpumalanga Evander Leslie Leandra 26-Nov-14 13-Mar-15
Mpumalanga Evander Embalenhle 26-Nov-14 13-Mar-15
Mpumalanga Witbank Delmas 27-Nov-14 9-Mar-15
Mpumalanga Witbank Witbank 28-Nov-14 9-Mar-15
KwaZulu-Natal Umlazi Umlazi 16-Jan-14 15-Jun-15
KwaZulu-Natal Umlazi Bhekithemba 16-Jan-15 15-Jun-15
KwaZulu-Natal Phoenix Inanda 17-Jan-15 16-Jun-15
113
KwaZulu
-Natal Phoenix Phoenix 18
-Jan
-14 16
-Jun
-15
KwaZulu
-Natal RK Khan Marianhill 19
-Jan
-14 17
-Jun
-15
KwaZulu
-Natal RK Khan Hillcrest 19-Jan
-15 17-Jun-15 
KwaZulu
-Natal Port Shepstone Port Shepstone 20
-Jan
-15 20
-Jun
-15
KwaZulu
-Natal Port Shepstone Hibberdene 20
-Jan
-15 20
-Jun
-15
KwaZulu
-Natal Stanger Umhlali 21
-Jan
-15 19
-Jun
-15
KwaZulu
-Natal Stanger Sundumbili 22-Jan
-15 19-Jun-15 
KwaZulu
-Natal Ngwelezane Esikhawini 23
-Jan
-15 18
-Jun
-15
KwaZulu
-Natal Ngwelezane Empangeni 24-Jan
-15 18-Jun-15 
KwaZulu
-Natal Edendale Taylors Halt 24
-Jan
-15 21
-Jun
-15
KwaZulu
-Natal Edendale Mountain Rise 25
-Jan
-15 21
-Jun
-15
Eastern Cape Taylor Bequest Afsondering 23
-Feb
-15 15
-Jul
-15
Eastern Cape Tayler Bequest Avondale 24
-Feb
-15 15
-Jul
-15
Eastern Cape Bizana Bizana 25
-Feb
-15 16
-Jul
-15
Eastern Cape Bizana Mpisi 26-Feb
-15 16-Jul-15
Eastern Cape Lusikisiki Hlababomvu 26
-Feb
-15 17
-Jul
-15
Eastern Cape Lusikisiki Lusikisiki 27
-Feb
-15 17
-Jul
-15
Eastern Cape Libode Libode 27
-Feb
-15 19
-Jul
-15
Eastern Cape Libode Ngqeleni 28
-Feb
-15 19
-Jul
-15
Eastern Cape Umthatha Umthatha 
1
-Mar
-15 20
-Jun
-15
Eastern Cape Umthatha Tsolo 1-Mar-15 18-Jul-15
Eastern Cape Butterworth Centane
2
-Mar
-15 21
-Jun
-15
Eastern Cape Butterworth Idutywa 
2
-Mar
-15 20
-Jun
-15
Eastern Cape Mdantsane Mdantsane
3
-Mar
-15 22
-Jun
-15
Eastern Cape Mdantsane East London 3-Mar-15 22-Jun-15 
Eastern Cape Port Elizabeth Motherwell
4
-Mar
-15 23
-Jun
-15
Eastern Cape Port Elizabeth Bethelsdorp 4-Mar-15 23-Jun-15 
Western Cape George George
5
-Mar
-15
6
-Sep
-15
Western Cape George Conville
5
-Mar
-15
6
-Sep
-15
Western Cape Worcester Worcester
6
-Mar
-15
7
-Sep
-15
Western Cape Worcester De Doorns
6
-Mar
-15
7
-Sep
-15
Western Cape Khayelitsha Harare
8
-Mar
-15
8
-Sep
-15
Western Cape Khayelitsha Lingelethu
-West 8-Mar-15 8-Sep-15 
Western Cape Bellville Kraaifontein *
8
-May
-15 N/A
Western Cape Bellville Delft * 10
-May
-15 N/A
Western Cape Mannenberg Gugulethu * 11
-May
-15 N/A
Western Cape Mannenberg Mitchells Plain *
9
-May
-15 N/A
North West Rustenburg Phokeng 12
-May
-15
1
-Sep
-15
North West Rustenburg Boitekong 12-May
-15 1-Sep-15 
North West Mafikeng Mafikeng 13
-May
-15
4
-Sep
-15
North West Mafikeng Madibogo 14
-May
-15
4
-Sep
-15
114
North West Potchefstroom Ventersdorp 14-May-15 3-Sep-15
North West Potchefstroom Klerkskraal 15-May-15 3-Sep-15
North West Klerksdorp Klerksdorp 15-May-15 2-Sep-15 
North West Klerksdorp Kanana 16-May-15 2-Sep-15
North West Taung Taung 30-Apr-15 5-Sep-15
Free State Sasolburg Deneysville 16-Apr-15 22-Sep-15
Free State Sasolburg Sasolburg 17-Apr-15 21-Sep-15
Free State Bethlehem Bethlehem 18-Apr-15 23-Sep-15
Free State Bethlehem Reitz 18-Apr-15 22-Sep-15
Free State Welkom Odendaalsrus 19-Apr-15 24-Sep-15
Free State Welkom Thabong 19-Apr-15 23-Sep-15
Free State Bloemfontein Bloemspruit 21-Apr-15 25-Sep-15
Free State Bloemfontein Park Road 21-Apr-15 24-Sep-15
Northern Cape Kimberley Kagisho 23-Apr-15 1-Oct-15
Northern Cape Kimberley Kimberley 23-Apr-15 1-Oct-15 
Northen Cape De Aar Britstown 24-Apr-15 2-Oct-15
Northern Cape De Aar Sunrise 24-Apr-15 2-Oct-15
Northern Cape Springbok Springbok 26-Apr-15 4-Oct-15
Northern Cape Steinkopf Steinkopf 26-Apr-15 4-Oct-15
Northern Cape Kuruman Bothithong 29-Apr-15 5-Oct-15
Northern Cape Kuruman Mothibistad 30-Apr-15 5-Oct-15 
Northern Cape Taung Pampierstat 1-May-15 6-Oct-15
115
ANNEX IX: COMPARISON OF MONTHLY CASES RECORDED THROUGH 
PRIMARY DATA COLLECTION AND MONTHLY CASES REPORTED BY THE 
NPA (OCT. 2015-SEPT. 2016)
As shown in the figure below, NPA utilization data is considerably higher in many cases than the primary 
data collected by the evaluation team directly from the TCCs. There are at least two potential 
explanations for these differences. First, the NPA data is from October 2015 to September 2016, whereas 
the evaluation data is from a selection of months between the fall of 2014 and the summer of 2015. 
Because these data span different time periods, we would not expect the numbers to completely match 
up. Implementers contend that utilization increased from 2015 to 2016. Furthermore, NPA data shows a 
seasonal drop in reporting during the second quarter, when most of our data collection occurred. Second, 
there is a possibility that one or both of the data sources are inaccurate. In the primary data collected by 
the evaluation, there is a possibility of under-reporting. This would contradict TCC statements from the 
24 TCCs for which we present primary data that the information provided is close to complete. There is 
also a possibility of over-reporting by the NPA, which would contradict NPA statements. For example, it 
is possible that the NPA data includes multiple visits from the same individual, while the evaluation data 
only includes the first visit of each presenting survivor. Does this affect the evaluation findings? It would 
certainly be better to have complete utilization data from the TCCs in which we feel confident and from 
which we can make descriptive inferences with more confidence. Nonetheless, as noted in the report, if 
there is a problem of underreporting, it should affect control and treatment precincts equally. As such, 
we feel that even provided the possibility of underreporting, the evaluation still offers an effective test of 
program impact. 
0
5
10
15
20
25
30
35
40
45
50
TCC
1
TCC
2
TCC
3
TCC
4
TCC
5
TCC
6
TCC
7
TCC
8
TCC
9
TCC
10
TCC
11
TCC
12
TCC
13
TCC
14
TCC
15
TCC
16
TCC
17
TCC
18
TCC
19
TCC
20
TCC
21
TCC
22
TCC
23
TCC
24
monthly cases - recorded (median of 12.7 cases per month)
monthly cases - NPA reported (median of 22.5 cases per month)
116
U.S. Agency for International Development
1300 Pennsylvania Avenue NW
Washington, DC 20523