DISCLAIMER This publication was produced for review by the United States Agency for International Development. It was prepared by NORC 
at the University of Chicago. 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.
IMPACT EVALUATION OF AN 
INTENSIVE JOURNALISMTRAINING
ACTIVITY IN TANZANIA
FINAL REPORT 
Prepared under Contract No. GS-10F-0033M / Order No. AID-OAA-M-13-00013, Tasking N061
DRG LEARNING, EVALUATION, AND
RESEARCH (DRG-LER) ACTIVITY
IMPACT EVALUATION OF AN
INTENSIVE JOURNALISM TRAINING
ACTIVITY IN TANZANIA
FINAL REPORT 
July 2021 
Prepared under Contract No.: GS-10F-0033M / AID-OAA-M-13-00013, Tasking N061 
Submitted to:
Mousumi Sarkar, COR
Submitted by:
Kristin Michelitch, Department of Political Science Vanderbilt University
Keith Weghorst, Department of Political Science Vanderbilt University
Contractor:
NORC at the University of Chicago
Attention: Matthew Parry, Program Manager
Bethesda, MD 20814 
Telephone: (301) 634-5444 
E-mail: Parry-Matthew@norc.org
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.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | ii
TABLE OF CONTENTS
ACRONYMS IV
EXECUTIVE SUMMARY V
EVALUATION DESIGN V
KEY FINDINGS VI
RECOMMENDATION FOR FUTURE PROGRAMMING AND EVALUATION VII
EVALUATION DESIGN RECOMMENDATIONS VII
IMPLEMENTATION RECOMMENDATIONS VII
1. INTRODUCTION 1
2. RESEARCH QUESTIONS AND THEORY OF CHANGE 2
RESEARCH QUESTIONS AND OUTCOMES 2
THEORY OF CHANGE 3
3. INTERVENTION DESCRIPTION 3
PILOT PROGRAM 3
BASELINE TECHNICAL ASSISTANCE FOR ALL STUDENTS 4
STUDENT JOURNALIST INTENSIVE TRAINING PROGRAM INTERVENTION 5
AUDIO PRODUCTION ACTIVITY 6
PROFESSIONAL WEBSITE 6
4. IMPACT EVALUATION DESIGN 6
RESEARCH DESIGN 6
RESEARCH DESIGN COMPONENTS 7
SCREENING SURVEY FOR GUAGING STUDENT INTEREST, AVAILABILITY, AND 
BACKGROUND 7
PILOT PROGRAM ASSESSMENT SURVEY 7
SECOND SCREENING SURVEY AND RANDOM ASSIGNMENT 8
BASELINE AND ENDLINE SURVEY 10
5. DATA COLLECTION 10
BASELINE DATA COLLECTION: 10
ENDLINE DATA COLLECTION: 11
6. FINDINGS 12
SUCCESS IN IMPLEMENTING BASELINE TECHNICAL ASSISTANCE 12
KNOWLEDGE OF PRACTICAL JOURNALISTIC PRACTICES 12
ETHICS AND GENDER INCLUSIVENESS 13
EFFICACY 14
TOPICAL INTERESTS FOR REPORTING 16
WERE PROGRAM EFFECTS DIFFERENT BY GENDER OR UNIVERSITY CONCENTRATION? 16
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | iii
7. DISCUSSIONS: REASONS FOR NULL EFFECTS AND VALIDITY 
CHALLENGES 17
INCONSISTENT INTERVENTION DELIVERY 17
ISSUES RELATED TO CURRICULUM DEVELOPMENT AND DELIVERY 17
STUDENT FEEDBACK WAS POSITIVE 17
SPILLOVER 18
NON-COMPLIANCE, ATTRITION, AND PARTICIPATION IN ALTERNATIVE PROGRAMS 18
SURVEY RESPONSE BIAS 20
STUDY LIMITATIONS 20
8. RECOMMENDATIONS 21
EVALUATION DESIGN RECOMMENDATIONS 22
IMPLEMENTATION RECOMMENDATIONS 22
APPENDIX A: SURVEY INSTRUMENT 24
APPENDIX B: REGRESSION RESULTS 53
LIST OF TABLES
Table 1. Smartphone and Audio Recording Capable Cell Phone Ownership, by University 5
Table 2. Recall and Application of Knowledge Outputs 8
Table 3. Sample Balance across Intervention, Control, and Waitlist on Major Characteristics at 
Random Assignment 10
Table 4. Knowledge of Accessing and Using University Resources 12
Table 5. Content and Production Knowledge 13
Table 6. Ethics and Gender Inclusion 13
Table 7. Efficacy 15
Table 8. Topical Interests for Reporting 16
LIST OF FIGURES
Figure 1. Impact Evaluation Theory of Change 3
Figure 2. Long-term Career Ambitions of Students (by Major) 4
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | iv
ACRONYMS
CSO Civil Society Organization
DRG-LER Democracy, Human Rights, and Governance Learning, Evaluation, and Research Activity
IAT Implicit Association Test
IE Impact Evaluation
IP Implementing Partner
NGO Non-governmental Organization
NORC National Opinion Research Center
RCT Randomized-controlled trial
SAUT Saint Augustine University of Tanzania
SJMC University of Dar es Salaam School of Journalism and Mass Communication
SMS Short Message Service
TMCS Tanzania Media and Civil Society Strengthening 
UDSM University of Dar es Salaam
USAID United States Agency for International Development
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | v
EXECUTIVE SUMMARY
This report presents the findings from an impact evaluation (IE) of an intensive journalism training
activity, which was designed through a collaborative process between the research team and USAID and
then implemented under the USAID/Tanzania Media and Civil Society Strengthening (TMCS) Program; 
the IE design and implementation was conducted by NORC at the University of Chicago (hereafter 
“NORC”) under the USAID Democracy, Human Rights, and Governance – Learning, Evaluation, and 
Research (DRG-LER) Activity contract. The activity aimed to train bachelor’s degree and diploma 
students, who were in media studies programs, especially those majoring in journalism or mass 
communications, as well as public relations, on practical skills in producing radio content, and to reduce 
gender, age, and rural identities bias that often persist in Tanzanian media. The intensive journalism 
training activity was conducted by Internews, TMCS’s implementing partner (IP). 
The current media environment in Tanzania is characterized by state repression of independent media, a 
climate of self-censorship, lack of equipment and professional deficiencies among journalists, limited 
journalistic independence, and gender, age, and rural identities bias. However, there are also critical 
opportunities that can be leveraged to strengthen the media environment, including that there are a 
significant number of civil society organizations (CSOs) focusing on introducing women’s and youth’s 
voices into the media. 
While many programs exist under the scope of USAID’s Strengthening the Media and Information 
Environment Initiative, the intervention evaluated in this report is an intensive journalism training 
program conducted through the “Media and Civil Society Strengthening (TMCS) Program”, known in 
Kiswahili as “Boresha habari”. The intensive training activity was designed to train students (a) in practical 
investigative journalism for audio production and (b) on diversity and inclusion of gender, age, and 
rural/urban groups to strengthen their ability to produce higher quality content as well as more diverse 
and attractive content regarding rural, youth, and women’s issues. The training activity had multiple 
learning units in this area. The training also included learning units on different news platforms, the role 
of internet and social media, how to edit and package a news story, and interviewing skills.
EVALUATION DESIGN
To evaluate the intensive journalism training activity, the NORC evaluation team launched a 
randomized-controlled trial (RCT). A total of 600 male and female Bachelor of Arts and Diploma 
students at accredited media studies programs (journalism, mass communication, and public relations) in 
the two academic institutions – University of Dar es Salaam School of Journalism and Mass 
Communication (SJMC) and Saint Augustine University of Tanzania (SAUT) in Mwanza were randomly 
assigned in equal numbers to the intervention group (i.e., those who received the program) or the 
control group. To do this, we first stratified the students by gender, year in the program, university, and 
academic major, and then conducted the random assignment from each stratum. The purpose of this 
stratified random assignment was to ensure that students in both the Intervention and control groups 
are as similar as possible based on the aforementioned characteristics.
Because the students of most interest were those majoring in journalism and mass communications 
(rather than public relations), when stratifying on major, we ensured all journalism and mass 
communication students were in the study either in the intervention or the control group. A majority of 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | vi
the students (60% in the intervention group and 59% in the control group) were journalism students and 
about 17% of the students were Mass Communication students. The public relations students “filled in 
the rest” of the slots for the two groups to achieve optimal comparability across the intervention (23%) 
and control (24% percent). 
All students in the study received baseline assistance such as availability of and knowledge to use 
recording equipment, computers, and editing software, which were provided through USAID program 
funding and delivered by the IP. We find that this was largely successful, although there is room for 
improvement for instructing students on accessing and using recording equipment.
To gather data for the IE, the evaluation team administered a baseline survey and an accompanying 
implicit association test (IAT) in August – October 2019. An endline survey with intervention and 
control groups was conducted in November 2019 – January 2020. The survey aimed to assess (1) 
knowledge of audio production and (2) diversity and inclusion. 
KEY FINDINGS
We find that the intensive training activity largely had a null effect on students’ knowledge of practical 
journalistic practices, as well as ethics and gender diversity and inclusiveness, which was overall low. 
Student efficacy in producing high quality journalism, a secondary outcome of interest, was also 
unaffected – most students with and without the program exposure expressed a high degree of efficacy. 
Those who undertook the program did, however, express more interest in covering topics that included 
women and rural issues, while those who did not have the program expressed more interest in sports 
and entertainment (interest in youth, politics and the economy were similar in both groups). We found 
no differences in program effectiveness by gender or concentration within media studies. 
Due to the COVID-19 pandemic, we were unable to engage students – both those who received and 
those who did not receive the training program – in a contest to produce radio segments. Originally, we 
planned to compare the quality of radio segments produced as another potential measure of program 
effects. The research team had been in discussions with the BBC (from the time of the scoping trip, and 
again after the endline), who expressed willingness to play the most successful segments produced, along 
with studio tours or other professional exposure for those students producing the successful segments. 
Further, the research team had planned to incentivize production by offering prizes (e.g., a tablet).
However, these discussions were still ongoing in February 2020; shortly thereafter, we needed to 
abandon planning for this part of the evaluation due to the pandemic. 
This report concludes with an assessment, informed by data, as to why there were largely null program 
effects. We largely rule out concerns of spillover, attrition, non-compliance, statistical power. Instead, 
the null effect could imply a number of issues. First, it is possible that the target population for the 
program was inappropriate. Some of the students in the media studies concentration eventually 
specialize in journalism or mass communications, while others end up specializing in public relations. 
While those in the later years who specialized in journalism and mass communications were prioritized 
for inclusion in the study (about 60% of the study sample), public relations concentrated students were 
also included to meet the minimum number of study participants required for the evaluation. It is 
possible that the null effects are related to the inclusions of these students. Second, the curriculum, as
was implemented, may not have been effective in having students retain knowledge after an intensive 
training program. Also, the curriculum was not standardized and different trainers were allowed to 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | vii
deliver the training differently. This was done with the intention of having trainers further develop the 
materials themselves. While such trainings may be effective in some contexts, this approach was 
problematic for the impact evaluation, which relied on a common training program for the purpose of 
the evaluation. Finally, the evaluation could not assess practical implementation of skills through a 
comparison of radio segments produced by students. Perhaps, even if knowledge did not stick for a “test 
format” akin to the online survey, for example, students in the program may have been more likely to 
participate in producing a radio segment, or produced a higher quality radio segment. Such a behavioral 
output is arguably a very good test of the program effect, and we regret not being able to implement it. 
RECOMMENDATION FOR FUTURE PROGRAMMING AND EVALUATION
We have identified a variety of potential causes for the program’s null effects, and recommend the 
following best practices and lessons when designing future research into similar programs. 
EVALUATION DESIGN RECOMMENDATIONS 
• We recommend USAID take a coordinated approach to add to its learning agenda through the 
implementation and evaluation of its activities. The most scientifically rigorous and useful learning for 
future USAID programming can be achieved by coordinating and aligning the goals of different 
stakeholders that contribute to a common learning agenda. Given the increased time demands on the 
IP when implementing an activity being evaluated through an RCT, expectations regarding staff time 
commitment, consistency of implementation, requirements for access to beneficiaries to conduct 
surveys and/or testing, time allotted throughout program implementation to account for required 
research tasks and similar considerations should be clearly defined and agreed upon by the funding 
agency, the research team, and IP at the start of the project.
• For IEs such as this one that are undertaken from a learning perspective (as opposed to an 
accountability perspective) to test the effectiveness of a concept, the intervention should be jointly 
developed by the research team, the IP, and USAID along with subject matter and local country 
experts who can tailor training materials to the specific concept and context being evaluated for the 
learning purpose. 
• Carefully consider access to technology in the local context when determining data collection 
strategies, and how it could serve as a barrier to participation. Despite the expansion of the internet 
in Tanzania, students had difficulty accessing the internet when they returned to their hometowns in 
the summer, which were often rural. IE designs need to include “defensive” measures to address 
such unanticipated issues, such as alternative modes of data collection or ensuring that key IE 
outcome measures are not solely dependent on a single data-collection mode.
IMPLEMENTATION RECOMMENDATIONS 
• All training material and data collection instruments should be translated into the local language for 
all phases of implementation. Where issues are encountered during implementation, this prevents 
assessment and resolution of the issue being complicated by ambiguity as to the source of the issue 
being within the training material/data collection instrument or language comprehension.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | viii
• Limit the amount of variability in intervention content, and administer the intervention consistently to 
all participants (i.e., ensure as much as possible that each beneficiary receives the identical 
intervention). Variability in the delivery of the intervention complicates analysis because differences in 
learning outcomes may be a result of the variability in the content of different training sessions. 
• To replicate successful programs, well-documented curriculum, implementation details, including 
deviation from plans and non-adherence to the plans, is absolutely necessary. Additionally, if a 
program yields null results, such documentation helps in understanding what may have been 
responsible for some of the null results. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 1
1. INTRODUCTION 
As part of the DRG Learning, Evaluation, and Research (DRG-LER) Activity, USAID requested that 
NORC design and implement an impact evaluation (IE) of “an intensive journalism training program” for 
university students in Tanzania, which was a part of USAID/Tanzania’s Media and Civil Society 
Strengthening (TMSC) Program. The multi-month intensive journalism training targeted bachelors and 
diploma students who were in journalism programs, especially those majoring in journalism or mass 
communications, as well as public relations, and aimed to strengthen their ability to produce higher
quality audio content. This intervention fits under USAID’s larger Strengthening the Media and 
Information Environment Initiative, which aims to increase the professionalization of journalists, connect 
the media to society, and strengthen the capacity of community radio. 
The importance of a rich, professional, and independent media environment is widely recognized as a 
key ingredient of a functioning democracy. The media can act as a vehicle that elicits and communicates 
citizen voices to the government and investigates and makes citizens aware of the performance of the 
government. The media can only fulfill such a function if it does not systematically exclude portions of 
the population, historically, women, youth, the poor, and rural populations. Following independence, 
Tanzania had few local journalists and the reporting environment was highly restricted under a one￾party regime. The media environment started to open in limited ways alongside the introduction of 
multiparty electoral competition during the 1990s and early 2000s. However, since the election of the 
recent government in 2016, the media environment has been closing. Prominent journalists have been 
harassed, tortured, and killed for criticizing the government, leading to a climate of self-censorship. 
Further, the ruling party has passed important pieces of legislation that restricts the activities of 
journalists. 
In Tanzania, the current media environment is further weakened by ongoing low professionalism, a 
product of poor training and incentives of media house owners. First, owners of media houses often 
employ those with little to no journalism training because they are cheaper to employ than trained 
journalists with a Diploma certification or higher. Yet, even Diploma and Bachelor of Arts students lack 
the practical training necessary to conduct high-quality journalism after degree conferral. Lack of 
equipment and opportunity for practical training plagues degree-granting institutions. Second, the media 
tends to focus on covering urban elites and media house editors can often be pressured to devote 
limited time or page-space to them. However, journalists, the vast majority of whom are free-lance, are 
poorly paid and do not have the means to travel to rural areas to cover ordinary citizens. Third, a 
patriarchal lens pervades, resulting in stories that degrade or scandalize women’s behavior. Women are 
often not interviewed or featured and issues of particular importance to women (e.g., gender-based 
violence) tend to be cast aside as unimportant. Similarly, patriarchal values regarding age hierarchy can 
silence youth voices and issues.
However, at this juncture there are also critical opportunities to strengthen the media environment. 
The Media Services Act requires that media content be produced by journalists with a minimum 
Diploma degree by 2021, which will force media house owners to employ and pay higher quality 
journalists. Second, women comprise the majority of student journalists in the high-ranking journalism 
programs, having the potential to increase the number of female journalists. Third, several civil society 
organizations focusing on increasing reporting on women’s and youth issues in the media have made 
large headway and gained backing from many media editors, the government, and religious stakeholders 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 2
in introducing women’s and youth’s voices and issues into the media, even sensitive issues such as sexual 
violence. Last, nationally-representative public opinion data show that Tanzanian citizens want to hear 
more about rural people, women’s issues, and youth issues. 
In response to these opportunities, USAID developed the Strengthening the Media and Information 
Environment Initiative to strengthen the quality of journalism, especially with regard to women and 
youth voices and issues. There are four components of the initiative: (1) strengthening the legislative and 
legal environment in which journalists may operate, (2) improving the professionalization of journalists, 
(3) fostering connection between the media and society (e.g., ordinary citizens and civil society), and (4) 
increasing the capacity of community radios. The medium of focus is radio, given the sheer dominance of 
radio – the vast majority of Tanzanians (78%) receive news over the radio at least a few times per week 
(versus 33% TV, 24% newspaper, and 9% Internet).
While many programs exist under the scope of USAID’s Strengthening the Media and Information 
Environment initiative, the intervention evaluated by this IE was conducted within the “Media and Civil 
Society Strengthening (TMCS) Program”, known in Kiswahili as “Boresha habari”. Under the TMCS 
program, an intensive training program was designed to train students in (a) practical investigative 
journalism for audio production and (b) diversity and inclusion of gender, age, and rural/urban groups to 
strengthen their ability to produce higher quality content as well as more diverse and attractive content 
regarding rural, youth, and women’s issues. 
2. RESEARCH QUESTIONS AND THEORY OF CHANGE
RESEARCH QUESTIONS AND OUTCOMES
The overarching research question and outcomes guiding this research are: What is the impact of audio 
investigative journalism and gender/youth/rural sensitivity training for young student journalists 
(Bachelor of Arts and Diploma track students)? We assess the following outcomes:
(1) Knowledge of Practical Journalistic Practices
(2) Ethics and Gender Diversity and Inclusiveness (the focus on rural and youth was dropped)
(3) Student Efficacy
(4) Topical Interests for Coverage
We also aimed at assessing short and/or mid-term career prospects (e.g., attaining internships/jobs; 
producing and submitting content to media outlets) for students who participated in the training 
program. However, due to lack of information on the students after they have graduated from the 
universities and the challenges related to tracking students during the COVID-19 pandemic in general, 
we were unable to assess this outcome. 
In addition, we had also aimed to assess the quality and appeal of radio segments produced by control 
versus intervention students in a contest (in collaboration with BBC). However, due to the COVID-19 
pandemic, we had to halt research activities before the contest could be launched. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 3
THEORY OF CHANGE
The theory of change for the professionalization of student journalists, around which this IE was 
designed, is in line with other types of vocational training programs (Figure 1). First, youth participation 
in high quality training should increase the capacity of participants to produce higher quality outputs. 
Second, production of higher quality outputs should increase the qualification of participants in the eyes 
of potential employers and thereby, improve career prospects. In the case of students continuing their 
education, they may also perform better on school metrics such as tests or practical assignments. Third, 
due to the gender/youth/rural sensitivity training component, the training may improve knowledge, 
attitudes, and behaviors regarding gender/youth/rural individuals on dimensions of diversity and 
inclusion. 
Longer term outcomes that are beyond the scope of the IE timeline (represented in dark red boxes) are 
the strengthening of journalistic content as the students advance in their careers and produce higher 
quality journalistic outputs. In turn, such a media environment should aid in empowering women, youth 
and rural folks, given that media is well-known to be agenda-setting and persuasive. 
Figure 1. Impact Evaluation Theory of Change
3. INTERVENTION DESCRIPTION
The intervention evaluated in this IE was implemented in several stages. The details of the intervention 
activity and the implementation stages were developed and finalized over several months of discussions 
between USAID DRG Learning Center, the USAID/Tanzania Mission, the IP, and the research team, 
including a scoping trip to Tanzania by the research team and a member of the DRG Learning Center. 
After the IE design was developed, the IP implemented a pilot to help understand the content and 
delivery of the training program. Then all students participating in the research study received technical 
assistance in audio recording and editing before the intensive training program was implemented. Finally, 
the IP implemented an intensive training program – the intervention – for randomly selected students in 
the participating universities. We discuss the details of the IE design in Section 4. Below, we discuss 
details of the intervention activities. 
PILOT PROGRAM
The pilot of the intensive training program was implemented in the summer of 2018 that took place in 
universities in Dar es Salaam, Mwanza, and Zanzibar; however, Zanzibar was not part of the IE as 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 4
implemented.1 For the pilot, the IP produced a draft training manual and installed several audio 
recorders and computers with audio editing software. The pilot program included both technical 
assistance in learning how to use the software, theoretical and practical training on how to produce high 
quality audio productions, and training on explicit gender bias. Students included in the pilot program 
were those who had already graduated from their respective universities. The research team conducted 
an informative endline survey assessment of this pilot program (pilot participants only) to gain more 
insight into the appropriateness of the curriculum and the delivery methods.
BASELINE TECHNICAL ASSISTANCE FOR ALL STUDENTS
In Spring of 2019, the IP equipped participating universities with recording devices and editing software 
for audio production. Students were then trained in successive sessions such that all bachelors and 
diploma students who were in media studies programs, especially those majoring in journalism or mass
communications, as well as public relations, received access and training to use this equipment to carry 
out audio recordings. At the training sessions for the software, students learned how to make a profile 
on a website that lists student journalists and allows them to list their resume details and to post their 
journalistic content.2 The IP built this website in January-February 2019. Based on a screening survey 
conducted to select students for the program, most participants expressed interest in long-term careers 
in journalism (Figure 2). This suggests that the professional website training has the potential to be useful 
for many students’ professional development.
Figure 2. Long-term Career Ambitions of Students by Major 
 
1 The IE did not include Zanzibar Journalism and Mass Media College because of the small student population expected to be 
enrolled in the Diploma and Bachelor of Arts programs related to journalism and mass communication.
2 This website will be hosted on the website of a partner organization, like the press club union. This roster of journalists would 
be open to ANY journalist or student journalist with a certificate or higher. It will thus build capacity in the media environment 
as part of overarching activities.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 5
Our screening survey, conducted before the baseline technical assistance (discussed in more detail in 
Section 4), showed that students generally had a device capable of recording audio (Table 1). However, 
in spite of students owning such devices, there were significant gaps in students’ ability and experience 
recording and producing audio content, as well as familiarity with audio recording and editing software. 
Thus, we viewed the baseline technical assistance both as a valuable service provided to the journalism 
students as well as a requirement to compare control and intervention students given common 
foundational skills (especially necessary, if we had been able to implement the radio segment contest). 
The baseline technical assistance was also designed to help pique interest in the intensive training 
program.
Table 1. Smartphone and Audio Recording Capable Cell Phone Ownership, 
by University
CELL PHONE TYPE UNIVERSITY YES NO
Smartphone
UDSM 23 (95.8%) 1 (4.2%)
SAUT 25 (71.4%) 10 (28.6%)
ZSJMCC 13 (61.9%) 8 (38.1%)
Audio Recording Capable Cell Phone
UDSM 21 (87.5%) 3 (12.5%)
SAUT 23 (71.9%) 9 (28.2%)
ZSJMCC 18 (90.0%) 2 (10.0%)
STUDENT JOURNALIST INTENSIVE TRAINING PROGRAM INTERVENTION
The intervention was an intensive training program that aimed to teach students to generate practical 
skills in producing radio content and reduce bias in reporting around gender, age, and rural identities, 
attitudes and behaviors that often persist in the Tanzanian media. The training program had multiple 
learning units in this area. The training also included learning units on different news platforms, the role 
of internet and social media, how to edit and package a news story, and interviewing skills. 
For the actual training program, the materials were translated to Swahili (as were the research team 
evaluation materials)3. Also, a component was added on implicit bias and the explicit bias section was 
modified as well.4
The training programs took place between semesters in the summer of 2019 at the University of Dar es 
Salaam (UDSM) and Saint Augustine University of Tanzania (SAUTI). Due to the number of students and 
difficulty locating and training enough trainers, the IP decided to hold two sessions per university. The IP 
 
3 The need to have training materials in Swahili even though universities in Tanzania use English as the teaching language 
stemmed from the feedback we received from the participants of the pilot training. We discuss those results in the next 
section.
4 However, a learning component on age and rural bias was not included in the final training materials. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 6
also had the challenge of fitting the training program around the academic schedules of the universities.
The dates for the two sessions were as follows:
– Session 1: July 22 - August 30, 2019
– Session 2: September 2 - October 11, 2019
AUDIO PRODUCTION ACTIVITY 
Universities and students received the aforementioned Baseline Technical Assistance (equipment and 
training) and Student Journalist Intensive Training Program Intervention conditional on the university 
facilitating the research activities (e.g., baseline survey, endline survey) and requiring that all students 
produce an original five to ten-minute audio segment investigating a topic of their choice. 
The student audio productions were supposed to be aired on community radio and BBC as part of an 
initiative to increase youth content on the radio. The best quality audio productions were also supposed 
to be highlighted at an event with media house editors and other stakeholders (e.g., TAMWA) with the 
possibility of receiving prizes. However, the evaluation of the student-produced audio production was 
cancelled due to the COVID-19 pandemic. If they had taken place, these events would have occurred in 
the summer of 2020 and would have been hosted by the IP and USAID. 
PROFESSIONAL WEBSITE
Another component of the TSMC activity was the creation of a professional website hosted by the IP 
that would serve as an electronic database for trained journalists in the country. The platform was 
intended to catalogue journalists living or working in Tanzania who possess a minimum required 
education level of a Certificate in Journalism from a verified institute. The website was also intended to 
provide a straightforward way for potential employers and partners to contact journalists, as the 
journalist profiles would contain their contact information. 
We aimed at assessing short and/or mid-term career prospects (e.g., attaining internships/jobs; 
producing and submitting content to media outlets) for students who participated in the training 
program based on information on this website. For example, one year after the program, we could 
assess, using data from this website, the number and quality of any content posted and jobs/internships. 
At the time of this analysis, the website was up and running but did not have sufficient content for 
career tracking. 
4. IMPACT EVALUATION DESIGN
RESEARCH DESIGN
We used a RCT to evaluate the intensive journalism training program conducted under the TMSC
program. The RCT was developed during the DRG IE Clinic held in Washington, DC March 27-31, 2017 
and on the scoping trip and follow-up activities in July and August of 2017. The population of interest 
was male and female diploma and bachelor students at accredited journalism programs in the two 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 7
academic institutions, University of Dar es Salaam SJMC and SAUT in Mwanza.5 In drawing the sample 
for the IE from this population of interest, we decided that the first priority would be journalism and 
mass communication students, then public relations students as a secondary priority. We discuss details 
of the IE design and the sample selection process below.
RESEARCH DESIGN COMPONENTS
The research team implemented two screening surveys (one in June 2018, one in May 2019), a pilot 
program assessment survey (following the pilot program conducted in the summer of 2018), a baseline 
survey (August – October 2019), and an endline survey (November 2019 – January 2020). The random 
assignment of students was done to create an intervention group, who received the intensive journalistic 
training program, and a control group who did not receive the intensive training. This assignment was 
conducted using the results of the second screening survey. 
SCREENING SURVEY FOR GUAGING STUDENT INTEREST, AVAILABILITY, AND BACKGROUND
The first screening survey was conducted in June 2018. The purpose of the survey was primarily to 
(1) collect information regarding student availability for the summer 2018 pilot program and (2) to 
identify general ownership of relevant equipment used during the training and for journalism activities. It 
also helped establish baseline expectations for student experience with digital media resources, interest 
in intensive training and, more generally, student interest in journalism topics and types. Results from 
this first screening survey showed that program interest in participation was high. 
PILOT PROGRAM ASSESSMENT SURVEY
A survey was conducted after the pilot intensive journalist training program in the summer of 2018. The 
survey was self-administered and conducted in English to match the English language used in the IP 
training manuals. This survey had multiple sections: (1) Basic Knowledge, (2) Future Plans and Subjective 
Program Impact, (3) Efficacy, and (4) Improvement Suggestions for Course Content and Instruction. 
In sum, although self-reported efficacy among students was very high following the training and 
instructors were lauded, students scored poorly on retention of knowledge and skills on both the 
training program’s content for radio production, as well as on diversity and inclusion. We elaborate on 
these results here.6
We assessed efficacy in terms of students’ belief on their ability to produce radio content and their 
perceptions of influence they might have through their content. Seventy-nine percent reported that they 
believed they would be completely able to independently produce a radio spot; 89% agreed or strongly 
agreed that they felt efficacy to do investigative journalism; 88% to understand political issues facing the 
country; and 99% to understand gender and youth issues facing the country. Virtually all (99%) believed 
they could do as good of a job as a journalist as their peers; 67% believed they have a say in what 
government does; and 44% believed that public officials care what journalists think.
 
5 The universities signed memoranda of understanding with the IPs. 
6 Please see research design document for a full set of results in Appendix C. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 8
Similarly, students rated the instructors highly – 85% of the students participating in the pilot rated 
instructor background in radio journalism and in diversity and inclusion as very or extremely 
knowledgeable. Students also rated instructors on several other categories on a 5-point scale from low 
to high with an average of 4.1 for enthusiasm, 3.5 for effectiveness in communicating ideas, 4.0 for 
receptiveness in answering student questions, 4.9 for encouragement of student participation.
However, students’ belief on their efficacy and the highly-rated instructors did not lead to high scores 
on knowledge retention. We included 12 questions of basic knowledge recall on audio production, 10 
vignettes (practical scenarios) requiring students to apply learning by identifying problems, and five 
questions on implicit bias.7 On average, students answered 33% of questions correctly and 67% 
incorrectly regarding recall of basic knowledge from the manual, were able to correctly apply knowledge 
of bias by identifying problems in practical vignette scenarios at a rate of 19% correct and 81% incorrect, 
and recall at a rate of 3% correct and 97% incorrect knowledge of implicit bias (Table 2). 
Table 2. Recall and Application of Knowledge Outputs
AVERAGE PERCENT
CORRECT INCORRECT
Basic Knowledge from Manual 33 67
Identifying Bias in Applied Vignettes 19 81
Knowledge of Implicit Bias from Circulated Materials 3 97
In open-ended questions for feedback, students remarked that the teacher should use Swahili only, 
which was evident in the English used by students to answer survey questions. Students thought that the 
ratio of student to equipment should be lowered, that stipends should be paid punctually, and the 
program should be lengthened. We presented the results and the feedback from the pilot in the design 
report and discussed incorporating changes based on these in the intervention training with USAID, the 
Mission, and the IP. 
SECOND SCREENING SURVEY AND RANDOM ASSIGNMENT
A second screening survey was administered in May 2019 to provide a more updated understanding of 
student availability and willingness to engage in intensive training for radio journalism in summer 2019. 
This survey attempted to reach all bachelors and diploma students who were in media studies programs, 
especially those majoring in journalism or mass communications, as well as public relations. There were 
993 respondents to this screening survey.
The data achieved two objectives: (1) assess receipt of the baseline technical assistance,8 and (2) gather 
“blocking covariates” for random assignment for invitation into the summer intensive training. Students 
 
7 Implicit bias was not included in the training materials. However, we circulated materials on implicit bias to the IP in Spring 
2018 that were incorporated into oral instructions. 
8 Although we planned to deliver the baseline technical assistance prior to the survey, the results indicated that not all students 
in the target population had received the baseline technical assistance. The IP delivered the baseline technical assistance to 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 9
were excluded from being eligible for the study if they were not interested and/or available for either of 
the two intensive training program sessions that would be offered as the intervention. Students were 
also excluded if they were not from one of the three majors. Of the 993 students, 941 were both 
interested and likely available for at least one of the two sessions.
Next, we randomly assigned students to the intervention versus control condition. The target sample 
size for the IE was 600 students – 300 for the intervention group and 300 for the control group – to 
ensure appropriate statistical power to detect any intervention effects. To do this, we first stratified the 
students by gender, year in the program, university, and academic major.9 We then conducted the 
random assignment within each stratum. This stratified random assignment or “block” random 
assignment ensured that both groups had a similar composition based on the above mentioned 
characteristics. We present the balance in the number of students in terms of these characteristics in 
Table 3. 
Because the students of most interest were those majoring in journalism and mass communications 
(rather than public relations), when blocking on major, we ensured all journalism and mass 
communication students were in the study either in the intervention or the control group. A majority of 
the students (60% in the intervention group and 59% in the control group) were journalism students and 
about 17% of the students were mass communication students (Table 3). The public relations students 
“filled in the rest” of the slots for the two groups to achieve optimal comparability across intervention 
(23%) and control (24%). 
Additionally, the research team provided the IP with a waitlist for students in the intervention group in 
case any drop outs occurred. The waitlist was numbered and was again meant to assure comparability 
across intervention and control groups. 
 
those who did not receive them, both to those randomly selected for the intensive program as well as the control group, 
before the intervention began.
9 We did not use GPA as a stratification variable because this was not available for first year students, who were the plurality of 
students. Furthermore, secondary school exam scores (e.g., in math, English) did not correlate well to GPA among the more 
senior students, and the exams were on sufficiently different topics. Thus, we could not use exam scores in a meaningful way to 
proxy for ability. However, random assignment should be fairly good at producing an even draw on ability across the two 
groups even without blocking on ability.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 10
Table 3. Sample Balance across Intervention, Control, and Waitlist on Major 
Characteristics at Random Assignment 
COVARIATES INTERVENTION 
GROUP
CONTROL
GROUP
INTERVENTION 
WAITLIST 
CONTROL 
WAITLIST 
Total 300 302 164 167
Gender
Male 174 177 86 93
Female 126 125 78 75
Major
Journalism 180 179 0 0 
Mass Communication 51 52 0 0 
Public Relations 69 71 164 168
Year
1 171 167 96 113
2 73 71 32 32
3 56 64 36 23
Finally, while many students were available for attending the training program at either session, some 
were only available to attend one of the sessions. For these students, we assigned them to the session 
for which they were available. For all other students, the session in which they participated was
randomly assigned. Furthermore, students were randomly assigned to instructors within the session. 
These random assignment procedures ensure that certain types of students do not end up together or 
with certain instructors. 
BASELINE AND ENDLINE SURVEY
Following the second screener survey, the team administered a baseline survey and an accompanying 
implicit association test (IAT) in August - October 2019. An endline survey with intervention and 
control groups was conducted in November 2019 - January 2020. The survey aimed to assess 
(1) knowledge of audio production and (2) diversity and inclusion. Both surveys were conducted using 
an online survey platform called Qualtrics. We provide details of the baseline and endline surveys in the 
next section. 
5. DATA COLLECTION 
BASELINE DATA COLLECTION
Baseline data collection was done using a survey followed by an IAT, both of which were to be 
administered in one sitting on a desktop computer. The survey and IAT instruments were first designed 
and translated, after which they were programmed in Qualtrics, a cloud-based online survey platform. 
Intervention students at both universities and in all sessions completed the survey and IAT in computer 
labs at their respective institutions. A total of 292 intervention students completed the survey and IAT 
across the two schools and two sessions. Due to logistical and timing issues, the data collection for the 
control group was conducted remotely at the end of the summer because there was no other 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 11
opportunity to have control students, who were not on campus during the summer, to complete the 
survey before returning to campus. Control group students needed to complete the survey before they 
returned to campus in order to avoid potential contamination in the form of information transmission 
from the summer training by intervention students to control group students when they would begin to 
intermingle at the start of the next semester of school. An incentive of 25,000 TZ shillings was provided 
to control students who completed the survey and the IAT in order to compensate for costs associated 
with taking the survey and IAT at internet cafes or other locations where they would be expected to 
pay since they did not have the option of using computer labs at the Universities, as the intervention
students were afforded. 
The research team faced considerable challenges in administering the survey to the control students 
remotely, which resulted in a low response rate. A majority of control group students in the research 
sample were not on campus, leading the research team to send the survey/IAT out via email. While the 
incentive provided to this population was intended to help offset some of the costs of time spent or 
travel to locations with reliable internet and desktop computers, engaging this population proved to be 
very challenging. The research team reached out to these students both by SMS and through phone calls 
in an effort to improve the response rate. However, the resulting response rate remained low and only 
22 percent of control students successfully completed the survey and IAT. The research team ultimately 
adjusted the research plan to accommodate an endline-only design, using lessons learned during the 
baseline to inform endline data collection.
ENDLINE DATA COLLECTION:
Endline data collection was conducted among both intervention and control groups between November 
2019 and January 2020. The instrument was similar in structure and content to the baseline instrument 
and included an online survey and IAT which needed to be completed in one sitting at a desktop 
computer. Students in both intervention and control groups were initially contacted by email which 
included a unique link to the survey and IAT. A local research assistant followed up with students 
through texts and phone calls to encourage participation. An incentive of 25,000 TZ shillings was 
provided to both intervention and control group students upon successful completion of both the 
survey and IAT. A total of 502 students participated: 257 from the control group and 245 from the 
intervention group, which resulted in response rates of 85.7% and 81.4% respectively.
As in baseline, the team had a difficult time convincing control students to take the endline survey, 
despite efforts to increase participation through mobilizing texts and emails by a local research assistant. 
Multiple contacts through both text and phone calls were made to each student who had not 
participated for several weeks in order to encourage participation. In checking for balance across these 
two groups, there is no difference in the proportion of men and women in intervention versus control. 
However, while the intervention and control groups have a similar proportion of mass communication 
students (17% and 19% respectively), the program group has a higher number of journalism students 
(57% versus 47% in the control) and a lower number of public relations students (26% versus 33% in the 
control). We discuss this imbalance further in Section 7 and the reasons why this ultimately was not an 
issue for the IE. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 12
6. FINDINGS 
We first describe results related to the successful delivery of the baseline technical assistance to all 
students. Next, we report the impact of the student journalism training, dividing into: (1) knowledge of 
practical journalistic practices, (2) ethics and gender inclusiveness, (3) efficacy, (4) topical interests, and 
(5) knowledge of university journalism resources. The question wording and protocols used for 
collecting data is presented in Appendix A, while detailed regression results are presented in Appendix 
B. In sum, we find that the program had little to no effect across these categories, which we elaborate 
below. In the next section, we explore possible reasons for the null to low effects and study limitations.
SUCCESS IN IMPLEMENTING BASELINE TECHNICAL ASSISTANCE
We asked students if they knew how to access and use university resources provided by USAID 
through the IP: recorders, computers, and editing software (Table 4). Since the goal of providing 
technical assistance was to reach control and intervention students equally, we should see high levels of 
knowledge and no differences across the control and intervention students. We see that, overall, 
knowledge of resources was very high for computers – at 98% and 99% of students in control and 
intervention groups. However, around a quarter of students in both groups did not have knowledge of 
how to access university recorders. Finally, while the students in the intervention group expressed very 
high levels of knowledge on using editing software, students without access to the activity (intervention) 
were significantly less likely to have such knowledge by eight percentage points.
KNOWLEDGE OF PRACTICAL JOURNALISTIC PRACTICES
Table 4. Knowledge of Accessing and Using University Resources
QUESTION CONTROL
(% AGREEING)
INTERVENTION
(% AGREEING)
Recorders 76% 76%
Computers 98% 99%
Edit Software 91% 99%
Using materials provided by the IP on practical aspects of radio segment creation, we crafted a series of 
questions that would display knowledge of the main points communicated in the materials. Specifically, 
we asked four questions on the creation of content, and two questions on production. Table 5 below 
lists the percentage of students who correctly answered questions among the intervention and the 
control students (results of statistical tests in Appendix B). Of the content questions, intervention
students generally had the same likelihood of getting answers correct as control students. In one case, 
the program had a statistically significant negative effect, although the negative effect is quite small in 
magnitude (Content Q3). There was similarly no change in likelihood of knowing the correct answer to 
production questions.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 13
Table 5. Content and Production Knowledge
QUESTION CONTROL
(% CORRECT)
INTERVENTION
(% CORRECT)
Content Q1: Writing Style 75% 78%
Content Q2: Central Principle of Writing for Audio 58% 68%
Content Q3: Number of Interviews 6% 4%
Content Q4: Definition of a “Cue” 8% 5%
Production Q1: Four Steps of Editing 29% 36%
Production Q2: Production Definitions 15% 20%
Note: Refer to Appendix A for full survey question wording.
ETHICS AND GENDER INCLUSIVENESS
Using materials provided by the IP, we crafted a series of five questions that would display knowledge of 
the main points communicated in the materials related to ethical aspects of journalism, including a 
learning module on gender balancing and issues of discrimination based on gender. Students in the 
intervention group were no more likely to correctly answer four of the questions, and they were 
statistically significantly less likely to correctly answer one question (Question 1); see Table 6 below.
Table 6. Ethics and Gender Inclusion
QUESTION CONTROL
(% CORRECT)
INTERVENTION
(% CORRECT)
Ethics Q1: Unconscious Bias 9% 6%
Ethics Q2: Affinity Bias 22% 29%
Ethics Q3: Confirmation Bias 27% 30%
Ethics Q4: Gender Equality Laws 34% 35%
Ethics Q5: Interviewing Youth 14% 13%
IAT D Score 0.2 0.2
Note: Refer to Appendix A for full survey question wording.
We also had students engage in an implicit association test (IAT). Currently, many scholars are 
researching whether trainings on discrimination towards marginalized groups can reduce negative 
implicit associations towards such groups. Implicit associations are automatic, unconscious biases that 
may emerge due to social norms and the media that lead people to associate certain people with certain 
traits. Implicit associations are thus beliefs that people may be unwilling to report because they are 
socially undesirable. However, implicit associations may also be something people are unable to report. 
For example, someone may believe that women and men should be equally associated with the public 
sphere, but may nonetheless automatically and unconsciously associate men with the public sphere to a 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 14
greater degree due to men’s historical and traditional role in the public sphere, or more frequent media 
portrayals of male leaders or men participating in the public sphere.
The IAT measures the strength of associations between a group of people (here, men and women) and 
stereotypes (here, public and private sphere). The IAT asks people to quickly sort words related to the 
categories of male and female (e.g., girl, male, boy, female) as well as public and private sphere (e.g., 
household, baraza (local discussion group meeting)) into categories on the left and right hand side of the 
computer or tablet screen. See Appendix A on the words used that were crafted to be locally 
appropriate. The IAT score is based on how long it takes a person, on average, to sort words together 
on the same side of the screen versus on opposite sides of the screen related to the groups and the 
stereotypes. Someone who can more easily sort words related to men and public sphere together than 
women and the public sphere together (and likewise, women and the private sphere versus men and the 
private sphere) would have an implicit association of men in the public sphere, and women in the private 
sphere. 
Many companies, government agencies (e.g., police), and universities are using implicit bias trainings in an 
effort to reduce biases, both explicit and implicit. However it is currently under investigation what types 
of trainings, specifically, might be effective. Thus, testing this program for effectiveness was an open 
social science and policy question.
The data show that students have implicit biases that dovetail with traditional stereotypes – they more 
readily associate men with the public sphere, and women with the private sphere. The intensive training 
program did not have a significant effect in reducing such a bias. These results are useful to 
understanding what trainings may not be effective at changing implicit biases.
EFFICACY
Students were asked a battery of questions regarding how confident they were in their abilities to 
undertake journalism. Students were asked about their strength in: (1) producing radio journalism, (2) 
gender sensitive reporting, (3) finding a story, (4) covering a story, and (5) producing a story. Students 
were also asked if they believed they were (6) qualified to perform investigative journalism, (7) could 
perform as a journalist better than most students, (8) could mobilize their community through their 
reporting, and (9) could identify fake news. Overall, students displayed a large degree of efficacy. While 
program beneficiaries often show higher efficacy, the difference between the intervention and control 
students is not statistically significant except for two answers (Efficacy Question 5 and 9). See Table 7
below.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 15
Table 7. Efficacy
QUESTION CONTROL
(% REPORTING)
INTERVENTION
(% REPORTING)
Efficacy Q1: Strength in Producing Radio Journalism
Poor: 0%
Average: 30%
Strong: 70%
Poor: 0%
Average: 22%
Strong: 78%
Efficacy Q2: Strength in Gender Sensitive Reporting
Poor: 1%
Average: 36%
Strong: 63%
Poor: 0%
Average: 32%
Strong: 68%
Efficacy Q3: Strength in Finding a Story
Poor: 0%
Average: 25%
Strong: 75%
Poor: 0%
Average: 24%
Strong: 79%
Efficacy Q4: Strength in Covering a Story
Poor: 0%
Average: 29%
Strong: 71%
Poor: 0%
Average: 24%
Strong: 76%
Efficacy Q5: Strength in Producing A Story
Poor: 1%
Average: 36%
Strong: 63%
Poor: 1%
Average: 22%
Strong: 77%
Efficacy Q6: Qualified Perform Investigative Journalism
Strongly Agree: 44%
Agree: 45%
Disagree: 9%
Strongly Disagree: 2%
Strongly Agree: 38%
Agree: 49%
Disagree: 12%
Strongly Disagree: 1%
Efficacy Q7: Perform as Journalist Better Than 
Other Students
Strongly Agree: 73%
Agree: 26%
Disagree: 0%
Strongly Disagree: 0%
Strongly Agree: 72%
Agree: 26%
Disagree: 2%
Strongly Disagree: 0%
Efficacy Q8: Mobilize My Community Through Reporting
Strongly Agree: 71%
Agree: 28%
Disagree: 0%
Strongly Disagree: 0%
Strongly Agree: 76%
Agree: 23%
Disagree: 0%
Strongly Disagree: 0%
Efficacy Q9: Identify Fake News
Strongly Agree: 63%
Agree: 33%
Disagree: 3%
Strongly Disagree: 1 %
Strongly Agree: 73%
Agree: 26%
Disagree: 1%
Strongly Disagree: 0%
Note: Refer to Appendix A for full survey question wording.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 16
TOPICAL INTERESTS FOR REPORTING
One aim of the student training was to increase interest in reporting on topics currently under 
represented in the media such as women’s, youth, and rural issues. Students were asked to name the 
topics that they were interested in reporting on the most. Students in the intervention group displayed 
significantly more interest in reporting on gender and rural issues, while students in the control group 
were significantly more interested in entertainment and sports.
Table 8. Topical Interests for Reporting
TOPIC CONTROL
(% REPORTING INTEREST)
INTERVENTION
(% REPORTING INTEREST)
Rural 46% 61%
Youth 69% 75%
Women 53% 67%
Sports 25% 20%
Politics 20% 16%
Economy 13% 8%
Entertainment 28% 18%
WERE PROGRAM EFFECTS DIFFERENT BY GENDER OR UNIVERSITY CONCENTRATION?
It may be that the program was differentially effective for males versus females, or for journalism versus 
mass communications versus public relations students. Perhaps null results were limited to one gender 
or particular majors, and the program may have been effective for other groups. For example, the IP 
expressed a concern that intervention effects would be smaller for public relations majors versus 
journalism or mass communications majors, even though all students had a basic foundation in media 
studies during the first two years of studies and all students were interested in the training that they 
were enrolled in. See Appendix B for regression results.
With regards to gender, we did not generally see statistically significant differences in the results.10
Similarly, we did not see many statistically significant differences across majors.11 While statistical power 
dwindles to assess statistical significance of small and medium effect sizes across these subgroups, we did 
not see any suggestive leanings that results would be different even with greater statistical power.
 
10 An exception is a positive treatment effect on Question 2.2 on production for females only.
11 One exception is that the program had a positive effect for Question 1.1 only among mass communications and public 
relations majors, but not journalism students. Similarly, QK3 only was more likely to be answered correctly by public relations 
students taking the program, versus journalism students (and the opposite for QK5). It also seems that, while the program 
buoyed interest in women’s issues and decreased interest in entertainment across all majors evenly, for public relations 
students, the program increased interest in rural areas (while decreasing it in sports and politics).
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 17
7. DISCUSSIONS: REASONS FOR NULL EFFECTS AND 
VALIDITY CHALLENGES
The evaluation suggests mainly null effects of the program. We consider multiple reasons, and also 
revisit some of the validity challenges proposed in the design document, namely, non-compliance with 
and attrition from the programs for intervention group students, and participation in alternative training 
programs by control group students. 
INCONSISTENT INTERVENTION DELIVERY 
ISSUES RELATED TO CURRICULUM DEVELOPMENT AND DELIVERY
One reason for null results may relate to issues regarding curriculum development and delivery. There 
were two primary concerns with the written training materials and its delivery: 
• First, there were different expectations regarding the training material development and delivery 
between the research team and the IP, but this difference didn’t emerge in the design/early 
implementation phase, but only once we were well into implementation. Training materials were not
fully developed and finalized prior to piloting. The training of trainers occurred with preliminary 
materials with the intention of having trainers further develop the materials themselves to encourage 
buy-in to the materials they would use to teach. This resulted in final training materials that did not 
include some aspects that the research team hypothesized to be necessary for the program to be 
effective; the IE was designed to empirically test that hypothesis. 
• Second, the lack of a standard curriculum resulted in trainers delivering somewhat different versions 
of what was intended to be a standardized radio production training. While such trainings may be 
effective in some contexts, it is highly problematic for an IE, which relies on evaluating a common 
intervention. In this case, each training was different. Accurate testing for program effects is 
complicated when different versions of the training are delivered to different students and there is no 
detailed documentation of the differences between sessions, as a result, it is difficult to define what is 
being evaluated.
STUDENT FEEDBACK WAS POSITIVE
In addition to assessing students’ knowledge, we also asked students to provide their own assessments. 
Those assessments were positive overall. However, it should be cautioned that students may feel 
compelled to provide more positive assessments if they believe doing so will make it more likely that a 
partnership with USAID/the IP would continue, either for themselves or for fellow students. 
We asked students who participated in the intensive training about the quality of the program in a few 
different areas. When asked about the pace of the class as fast or slow, 83% rated the class as about 
right, with 12% saying it was a little fast and 5% saying it was much too fast. When asked about the 
instructor’s knowledge, responses were moderate. When asked about their instructor’s knowledge of 
media, 23% said the instructor was very knowledgeable, while 72% rated them in the middle, and 4% as 
low. Instructors were also rated in the middle regarding their knowledge of inclusion and diversity: 17% 
reported high knowledge, 73% in the middle, and 9% low. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 18
However, instructors received quite a high rating on other teaching dimensions. Regarding their 
enthusiasm with the material – 73% of students rate their instructors as very good, and 23% as good. 
Regarding their effectiveness in communicating ideas, 81% rated instructors as very good and 17% as 
good. On their receptiveness to student questions, 77% rated instructors as very good with 18% giving 
the rating of good; 86% rated instructors as very good in encouraging students to speak in class, with 
10% rating them as good. Equal treatment by the instructor for female and male students was also rated 
at 83% very good and 13% good. 
SPILLOVER
Spillover occurs when students in the control group receive the intervention in a secondhand fashion 
from intervention group students. Given that no control students were able to interlope on the 
intervention activity (they were home during the semester break), first hand receipt of the intervention
is of little concern. However, students who received the program may have passed on knowledge to 
students that were not randomly selected to receive the program after the university was back in 
session. 
On the survey we asked students who did NOT have the program whether they talked to students in 
the training program about the training content. Spillover could be a concern, since 31% of them talked 
“many times” and 28% “sometimes” while only 21% reported talking only “a few times” and 20% 
reported “never.” Of course, we don’t know what they talked about or the extent of discussion. 
Taken together, we have little worry about spillover because many of the responses for both 
intervention and control reveal low levels of knowledge rather than high levels of knowledge. When 
correct answers across the intervention and control groups are both equally low, then perhaps the 
training was not effective. However, where correct answers across the two groups are both equally 
high, spillover may be an issue.
NON-COMPLIANCE, ATTRITION, AND PARTICIPATION IN ALTERNATIVE PROGRAMS
We also assess a number of issues that may cause imbalances across intervention and control groups, in 
terms of student background characteristics, that confound the intervention assignment, and explain null 
results. These issues are related to non-compliance, student attrition from the program, and 
participation in alternative programs by control students.
To address non-compliance due to financial issues (for example, the screening survey identified that a 
number of students were unable to afford housing between semesters to participate) the evaluation 
team and the IP ensured that students were given payment for participation to cover such needs. 
However, things can always come up between the time that students project that they could participate 
and the time that the program takes place. In order to ward off potential imbalances, we created a 
waitlist for the program and control groups that would ensure balance. Only 13 students in the 
intervention dropped out prior to the training and were replaced with waitlist students (three in SJMC 
and ten in SAUT).
However, one issue that arose is that some students enrolled in the intensive training program had to 
retake final exams and missed initial program days of the first session. The IP was not aware that this 
might be potentially a time conflict when they scheduled the training days. Detailed documentation on 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 19
student absences and on which students were paid sitting allowances for which training days was not 
available from the IP, so we could not assess to what extent it affected the impact results. If this 
problem of missing initial training days were severe, it might explain null results. However, in many 
cases, the differences between control and intervention are not close to significant, and intervention
group levels (as well as control group) of correct responses are very low. Thus, we are not very 
concerned that this issue is critical. 
A second concern may be that there was survey attrition in the endline that produced an imbalance 
(e.g., the best students in the intervention group did not happen to take the survey while the best 
students in the control group did), and that this imbalance is responsible for lack of results. There were 
six individuals who took the survey (five in control, one in the intervention) who were not in the 
original population assessed in the screening survey that was used to generate intervention and control 
groups (or at least, could not be matched to individuals in that population). In the end, 82% of the 
original group assigned to the intervention took the endline survey (either because they did not end up 
taking the program or they took the program and did not answer the endline survey). Note that, due to 
13 students assigned to the intervention dropping out of the program, they were replaced with 13 
intervention wait list individuals, of whom 12 took the endline survey). Of the original control group, 
71% took the endline survey, and the rest of the survey takers were recruited from the control wait list. 
The reason for this recruiting was to ensure balance across intervention and control. 
We checked the balance between those who completed the surveys in the intervention group and those 
who completed the survey in the control group to ensure that the two groups were still comparable. 
We found balance was achieved on gender and there was a slight imbalance where there were more 
journalism and less public relations students proportionally in the intervention group versus the control. 
However, we are not concerned about the latter imbalance driving results because we examined 
heterogeneous intervention effects by major concentration, and we found null results in both average 
intervention effects as well as heterogeneous intervention effects. Because we do not have GPA for 
most of the students, we cannot assess imbalance in the endline survey on ability. In sum, we are not 
particularly concerned with the issue of attrition affecting study conclusions.
A third and related concern for the study is whether those not randomly selected for the intensive 
training had access to alternative trainings or experiences that boosted skills. Having access to 
alternative, similar training may, however, only explain null intervention effects for those few areas 
where there were high levels of knowledge or efficacy across both groups. 
Indeed, it appears control group students did have some exposure to alternative trainings or may have 
engaged in some journalism activities on their own. Many students in the control group reported 
producing an audio segment with others in a group – 51%, while 52% reported producing one on their 
own. Around a third (34%) said they received training on producing an audio segment and around a fifth 
(21%) received training on gender/youth inclusion. Seventeen percent reported having an internship 
(whereby 58% of those with an internship had one related to media) and 24% had a job (whereby 60% 
of those with a job had one related to media). Of course, many helped their family with agriculture, 
household, or business tasks – 53% – and 25% undertook leisure. Thus, it seems that some students may 
have undertaken practical production of audio segments on their own, outside the context of an 
internship or a job. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 20
SURVEY RESPONSE BIAS
Respondents could falsely report on a survey with an alternative goal. This is not a concern with 
knowledge assessment or the IAT, but could be a concern for the questions on efficacy or assessment of 
the program. For example, students could report being overly confident regarding their own abilities in 
both intervention and control groups, and/or treated students could report being overly satisfied with 
the program if they believe that the answers may affect future eligibility for a program access or USAID 
partnership with their university program. Of course, student efficacy could just be high from university 
training and the intensive program does little to boost such high efficacy. 
STUDY LIMITATIONS
This evaluation has a few limitations that we have briefly touched on elsewhere in the report, but we 
summarize them here. 
• First, one limitation of this study is that initially the design called for an assessment of participation in 
a five minute radio segment contest. The research team had been in discussions with the BBC (from 
the time of the scoping trip, and again after the endline), who expressed willingness to play the most 
successful segments produced, along with studio tours or other professional exposure for those 
students producing the successful segments. Further, the research team had planned to incentivize 
production by offering prizes (e.g., a tablet). However, these discussions were still ongoing in 
February 2020 and given the COVID-19 pandemic, we needed to abandon planning for this part of 
the evaluation. Carrying out such a contest would have put students in direct contact with others 
and increased public risk of disease. By examining who would produce a segment, as well as the 
quality of the segment, the evaluation could have assessed practical implementation of skills. Perhaps, 
even if knowledge did not stick for a “test format” akin to the online survey, for example, students in 
the program may have been more likely to participate in producing a segment, or produced a higher 
quality segment.12 Such a behavioral output is arguably a very good test of the program effect, and we 
regret not being able to implement it. 
• Second, while we have statistical power to detect moderate to large program effects, statistical 
power is limited in detecting small program effects (as noted in the design document) due to a limited 
number of treated students. However, we note that predicted probabilities and values are very 
similar across intervention and control groups and confidence intervals are not particularly similar. 
This means, if we had more statistical power (i.e., more students enrolled in the program and a 
commensurate increase in control students), it is unlikely that differences would emerge and reach 
statistical significance. We note that we present results without clustering standard errors at the 
classroom level, since this would further limit statistical power and be even less likely to reveal 
program effects. 
• Third and relatedly, the original design called for a difference-in-difference evaluation to enhance 
statistical power. However, due to logistical and timing issues, the data collection for the control 
group was conducted remotely at the end of the summer because there was no other opportunity to 
 
12 One issue we noted, however, is universities seemed reluctant to require students to create a segment, so one limitation of 
the analysis may have been that it is hard to compare content across control and treatment if only certain people, or different 
types of people, select into production of a segment in each group.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 21
have control students, who were not on campus during the summer, to complete the survey before 
returning to campus. The research team faced considerable challenges in administering the survey to 
the control students remotely and 70% of the control group did not complete the baseline survey 
which was insufficient to conduct a difference-in-difference analysis. The reasons students gave for 
not taking the baseline survey were as follows: 52% said they did not receive an email requesting 
participation, 26% said they did not get an SMS message asking them to participate, 36% said they had 
no access to a computer, and 14% had no access to the internet. Only 2% said they meant to but 
forgot. None said that the reason was inadequate payment. We do note that failure to get prompt 
payment, due to technical difficulties of transferring money internationally at high speed, angered 
some students in previous data collection efforts.
• Fourth, one issue that was of special concern to the IP was that students in the target population 
were inappropriate. Some of the students in the media studies concentration eventually specialize in 
journalism or mass communications, while others end up specializing in public relations. While those 
in the later years who specialized in journalism and mass communications were prioritized for 
inclusion in the study, public relations concentrated students were also included to meet the 
minimum number of study participants required for the RCT (both for intervention and control, 
since groups must be balanced). 
The IP preferred to train small-groups of highly-motivated students working in university radio 
stations or on university newspapers, and mentor them over a couple of years to have them produce 
better radio contents with the equipment made available to them; this was one of the typical 
populations to whom the IP had delivered its capacity-building (i.e., highly focused on journalists in the 
traditional sense). While such models were used in previous projects, it is extremely difficult to 
address the selection bias in evaluating such programs. Also, given USAID’s interest in evaluating an 
intensive training program for general students who may eventually choose a career in journalism and 
the need for the IE to include a large number of students to ensure statistical power, the IE included 
students in journalism, mass communication, and public relations. Although journalism students were 
prioritized, the null impacts could be related to the inappropriate student population as pointed out 
by the IP. 
The research team felt the inclusion of these students would not significantly impact the evaluation 
for a variety of reasons. First, all students who are in the study (treatment and control) expressed 
interest in program enrolment. Second, students had common foundational coursework from earlier 
years and were all enrolled in media studies. Third, due to job incentives, some students who had 
been planning to follow the public relations path may still change to become journalists as they 
continue through school. This means the students’ declared major concentration might not reflect 
actual career interests and the intensive program may enable some students to obtain internships or 
jobs in journalism who may have formerly believed that that was not possible.
8. RECOMMENDATIONS
The impact evaluation was unable to detect a measurable impact from the program. We have identified 
a variety of potential causes for these null effects and recommend the following best practices based on 
the lessons learnt when designing future research into similar programs.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 22
EVALUATION DESIGN RECOMMENDATIONS
• Include specific requirements of the RCT in the agreement with the IP to ensure the IP’s program 
will accommodate the IE research design requirements. 
− This should include the early establishment of relationships with partner organizations, (such as 
Universities in this instance), with whom both the researchers and the IP need cooperation from 
the onset of design. In retrospect, the research team would have benefitted from more direct 
communication with the Universities (rather than having all communication about the IE pass 
through the IP); this might have ensured they were more well-informed about aspects of the 
intervention that were of greater importance for them and quickly or more completely uncovered 
some of the complications that were discovered during the IE implementation. Having partner 
organizations whose role is critical to the success of the intervention design and implementation 
fully informed is crucial to the success of the IE. 
− Given the increased time demands on the IP when implementing an activity being evaluated 
through an RCT, expectations regarding staff time commitment, consistency of implementation, 
requirements for access to beneficiaries to conduct surveys and/or testing, time allotted 
throughout program implementation to account for required research tasks and similar 
considerations should be clearly defined and agreed upon by the funding agency, the research 
team, and IP at the start of the project. Although the research team believed it had fully conveyed 
to the IP what was involved in conducting the IE, there were many instances where the IP did 
not fully appreciate the effort that was required to fulfill the research design until the task was 
under way.
• For IEs such as this one that are undertaken from a learning perspective (as opposed to an 
accountability perspective) to test the effectiveness of a concept, the intervention should be jointly 
developed by the research team, IP, and USAID along with topic and local country experts who can 
tailor training materials to the specific context and the concept being evaluated for the learning 
purpose. 
• Despite advances in technology and the expansion of internet in recent years in Tanzania, there 
remain significant challenges with access to technology which impacted the ability of control group 
students to effectively participate in the survey once they had left campus for the summer and 
returned to their hometowns, which were often rural and did not afford easy, reliable access to the 
internet or computers. The IAT required such a connection and could only be completed on a 
computer (rather than a smartphone) which complicated participation for several students. The 
original IE design anticipated all data collection would occur on campus, where students would have 
easy access to adequate computing resources. IE designs need to include “defensive” measures to 
address such unanticipated issues, such as alternative modes of data collection or, in the case of a 
mode-specific data collection like the IAT, ensuring that key IE outcome measures do not have 
such dependencies.
IMPLEMENTATION RECOMMENDATIONS
• All training materials should be translated into the local language for all phases of implementation. For 
this program, the pilot relied upon draft versions of English language materials rather than finalized 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 23
versions translated into Swahili. This complicated the research team’s ability to learn from the pilot; it 
was difficult to determine if observed deficiencies in pilot outcomes were due to the materials being 
used or if pilot participants simply did not understand the materials being used due to language 
barriers. The training materials were later translated into Swahili and revised prior to use during 
implementation of the main summer program, but the opportunity to use the pilot to improve the 
quality and effectiveness of the training materials and update them accordingly was limited. 
• To fully understand what is being evaluated, it is important for all materials used to implement the 
training be well documented and administered consistently to all participants (i.e., ensure as much as 
possible that each beneficiary receives the identical intervention). This means that similar programs 
being evaluated through RCTs in the future should aim to limit the amount of variability in content 
and topics covered across multiple training sessions. Content, including examples of key concepts, 
within some training modules was developed by the students and teachers through conversations and 
brainstorming sessions. This complicates analysis because differences in learning outcomes may be a 
result of the variability in the content of different training sessions. 
• To replicate successful programs, well-documented curriculum, implementation details, including 
deviation from plans and non-adherence to the plans, is absolutely necessary. Additionally, if a 
program yields null results, such documentation helps in understanding what may have been 
responsible for some of the null results. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 24
APPENDIX A: SURVEY INSTRUMENT
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 25
ENDLINE - DRG TANZANIA IE
Start of Block: CONSENT BLOCK
This is a short survey geared towards students in tertiary degrees for journalism, mass communication, 
and public relations. The survey covers aspects of media production, as well as the media more generally 
in Tanzania. This study will use the Implicit Association Test (IAT) as well, a specific type of survey that 
will be described later. The survey should not take more than 45 minutes. It is important to answer 
questions as honestly as possible, with as little outside distraction as possible. The results of the study 
will inform future student programming. Upon completion of the survey, you will receive 25,000 
shillings. This money will be sent through your mobile phone provider (Mpesa, Tigopesa, AirtelMoney, 
etc.) 
Your responses are confidential. Only the researchers from NORC will have access to the data. The 
information you and others provide will be summarized in a report. Your name or any other identifying 
information will never be used when we report results of the survey. Your participation in this survey is 
completely voluntary – it is your choice to take this survey or not. If you feel uneasy with any of the 
questions, you may skip those questions without answering them. You may also stop the survey at any
time. If you decide not to take part or to stop the interview, you will not lose any services that you are 
otherwise receiving. 
By clicking the ‘Proceed’ button at the bottom right of the screen, you acknowledge your acceptance to 
participate in this survey. Thank you very much for your participation. 
End of Block: CONSENT BLOCK
Start of Block: MODULE 0: FIELD CONTROL
first_name Please type your first name
location Select your location:
• Mwanza (1) 
• Dar es Salaam (2)
student_ID_1 Please enter your student ID:
student_ID_2 Please re-enter your student ID:
mobile_1 After completing this survey, you will be sent a payment through your mobile 
phone provider. Please enter your phone number so we can send you this payment:
mobile_2 Please re-enter your phone number so we are able to send you this payment:
End of Block: MODULE 0: FIELD CONTROL
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 26
Q1_1 In the “inverted pyramid style” of writing, the most important information is 
found…
o At the top (1)
o In the middle (2)
o At the bottom (3)
Q1_2 The central principle of writing for radio is… 
o Following editorial instructions
o Your passion makes good stories
o The listener comes first
o Avoiding “dead air” at all costs
Q1_3 Generally speaking, you should interview about how many people for 5-6 good vox 
pops? 
o One is enough (1)
o 2-4 (2)
o 5-7 (3)
o 8-10 (4)
o 11-15 (5)
o 15-20 (6)
Q1_4 A “cue” should… (select all that apply)
o Summarize what is to follow 
o Give a listener a sense of anticipation 
o Be less than 30 seconds long
o Repeat opening words of the report
Q_2_1 T The four steps of editing are: (a) Mix, (b) Edit and start your cue sheet (c) Listen 
to voice elements, and (d) Listen to non-voice other elements. 
Please order these steps using numbers 1 through 4 to indicate which is the first step; the 
second step; the third step, and the fourth step. 
_____ Mix (1) 
_____ Edit and create a cue sheet (2)
_____ Listen to vocal components (3)
_____ Listen to non-vocal components (4) 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 27
Q_2_2 Please select the word that matches the definitions that follow.
Q_2_2_0_GROUP (a) A short song or piece of music, which acts an audio logo or brand.
o Actuality (1)
o Wild Track (2)
o Sounder (3)
o Jingle-jingle (4)
Q_2_2_1_GROUP (b) Natural sound of the location.
o Actuality (1)
o Wild Track (2)
o Sounder (3)
o Jingle-jingle (4)
Q_2_2_2_GROUP (c) The recording of something happening.
o Actuality (1)
o Wild Track (2)
o Sounder (3)
o Jingle-jingle (4)
Q_2_2_3_GROUP (d) A recorded tune used to introduce segments of the broadcast.
o Actuality (1)
o Wild Track (2)
o Sounder (3)
o Jingle-jingle (4)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 28
intro_experiments In the following section, we are going to present to you stories that you 
might find in the media in Tanzania. We would like for you to read each story carefully. 
After each, you will be asked your opinion about the contents of the story. You will also be 
asked to recall some basic facts from the article. 
Q116 Please read the following hypothetical news story on fake news law. You will be 
asked your opinion about the contents of the story. You will also be asked to recall some 
basic facts from the article. 
Randomized to QE_fnccmpres, QE_fnccmparty, QE_fnmixpres,QE_fnmixparty_text
QE_fnccmpres
Headline: New Commission—appointed by President Magufuli—Recommends increased penalties for 
spreading “fake news” about the government 
A new government commission has been created in Tanzania to reduce the spread of misinformation/ 
“fake news.” The Commission has proposed amending the Media Services Act to include stricter 
penalties for those accused of disseminating misinformation/ “fake news.” It will allow the commission 
to arrest and imprison journalists and editors accused of violating the law. 
Supporters of the proposal claim that severe penalties are needed to combat misinformation/”fake 
news” about the country’s economy and politics. Those who oppose the proposal claim that 
the commission may use its power to target critics of the current government and to restrict of 
freedom of expression. The members of the commission - who are CCM members - were all appointed 
by the President and their pictures are found above. After careful scrutiny, the commission 
(pictured above) recommended to increase penalties for spreading misinformation/ “fake 
news” about the government. 
QE_fnccmparty
Headline: New Commission—appointed by the ruling party—Recommends increased penalties for 
spreading “fake news” about the government 
A new government commission has been created in Tanzania to reduce the spread of misinformation/ 
”fake news.” The Commission has proposed amending the Media Services Act to include stricter 
penalties for those accused of disseminating misinformation/ ”fake news.” It will allow the commission 
to arrest and imprison journalists and editors accused of violating the law. 
Supporters of the proposal claim that severe penalties are needed to combat misinformation/ ”fake 
news” about the country’s economy and politics. Those who oppose the proposal claim that 
the commission may use its power to target critics of the current government and to restrict of 
freedom of expression. The members of the commission - who are CCM members - were appointed 
by their political parties and their pictures are found above. After careful scrutiny, the commission 
(pictured above) recommended to increase penalties for spreading misinformation/ “fake 
news” about the government. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 29
QE_fnmixpres
Headline: New Commission—appointed by President Magufuli—Recommends increased penalties for 
spreading “fake news” about the government 
A new government commission has been created in Tanzania to reduce the spread of misinformation/ 
”fake news.” The Commission has proposed amending the Media Services Act to include stricter 
penalties for those accused of disseminating misinformation/ ”fake news.” It will allow the commission 
to arrest and imprison journalists and editors accused of violating the law. 
Supporters of the proposal claim that severe penalties are needed to combat misinformation/”fake 
news” about the country’s economy and politics. Those who oppose the proposal claim that 
the commission may use its power to target critics of the current government and to restrict of 
freedom of expression. The members of the commission - who are members of various political parties 
- were all appointed by the President and their pictures are found above. After careful scrutiny, the 
commission (pictured above) recommended to increase penalties for spreading 
misinformation/ “fake news” about the government. 
QE_fnmixparty
Headline: New Commission—appointed by political parties—Recommends increased penalties for 
spreading “fake news” about the government 
A new government commission has been created in Tanzania to reduce the spread of misinformation/ 
”fake news.” The Commission has proposed amending the Media Services Act to include stricter 
penalties for those accused of disseminating misinformation/ ”fake news.” It will allow the commission 
to arrest and imprison journalists and editors accused of violating the law. 
Supporters of the proposal claim that severe penalties are needed to combat misinformation/ ”fake 
news” about the country’s economy and politics. Those who oppose the proposal claim that 
the commission may use its power to target critics of the current government and to restrict of 
freedom of expression. The members of the commission - who are members of various political parties 
- were appointed by their political parties and their pictures are found above. After careful scrutiny, 
the commission (pictured above) recommended to increase penalties for spreading 
misinformation/ “fake news” about the government. 
QH_1 In the article about “Fake News,” among the eight committee members, how many 
CCM members were on the committee that considered the policy about “fake news?”
QH_2 In an article on “Fake News”, what decision did the committee make regarding 
punishment for disseminating “Fake news?” 
o Increasing penalties for disseminating “Fake News” (1) 
o Decreasing Penalties for disseminating “Fake News” (2) 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 30
QH_3 In the article on “Fake News,” who were the committee members selected by? 
o President Magufuli (1)
o Their political parties (2) 
Q_E_1 Do you agree or disagree with the following statement: Personally, I think the 
committee made the right decision. 
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_2 Do you agree or disagree with the following statement: This decision is for the 
better of the media environment in Tanzania. 
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_3 Do you think decision-making process was fair?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_6 Do you think the committee can be trusted to make decisions that improve the 
efficiency of government in Tanzania?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 31
Q_E_7 Thinking about the way the committee was chosen, do you agree or disagree that 
the committee can make decisions without influence from political actors?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_8 Thinking about the partisanship of committee members, do you agree or disagree 
that the committee can make decisions without influence from political actors?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_9 Thinking about the way the committee was chosen, do you agree or disagree that 
the committee can make decisions that represent Tanzanians in general?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_10 Thinking about the partisanship of committee members, do you agree or disagree 
that the committee can make decisions that represent Tanzanians in general?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 32
Q_E_11 Thinking about the way the committee was chosen, do you agree or disagree that 
the committee can make decisions that improve the quality of democracy in Tanzania?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_E_12 Thinking about the partisanship of committee members, do you agree or disagree 
that the committee can make decisions that improve the quality of democracy in 
Tanzania?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q117
Please read the following hypothetical news story about a sexual harassment law. You will 
be asked your opinion about the contents of the story. You will also be asked to recall 
some basic facts from the article. 
Option 1: Mixed panel – increase (harassment_expand_mixedgender.jpg)
The Arusha City Council formed a special committee to consider legislation that would increase 
penalties for sexual harassment in the workplace. The law would require businesses to discipline 
employees found guilty of sexual harassment. Penalties could include immediate suspension or 
termination. 
The law would use a standard definition of sexual harassment, which includes unwelcome sexual 
advances, requests for sexual favors, and other verbal or physical harassment of a sexual nature. In 
Tanzania, most sexual harassment complaints are filed by women. 
Supporters claim the increase penalties will help end sexual harassment in the workplace. Those 
opposed to the law claim the law will excessively punish misunderstandings between employees. 
After careful discussion, the committee (pictured above) voted to increase penalties for sexual 
harassment. 
Option 2: Mixed panel – decrease (harassment_reduce_mixedgender.jpg)
The Arusha City Council formed a special committee to consider legislation that would decrease 
penalties for sexual harassment in the workplace. The law would restrict business’ ability to discipline 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 33
employees found guilty of sexual harassment. Penalties could not include immediate suspension or 
termination of employment. 
The law would use a standard definition of sexual harassment, which includes unwelcome sexual 
advances, requests for sexual favors, and other verbal or physical harassment of a sexual nature. In 
Tanzania, most sexual harassment complaints are filed by women. 
Supporters claim the decreased penalties will protect employees who have experienced 
misunderstandings at work. Those opposed to the law claim the law will increase sexual harassment in 
the workplace. 
After careful discussion, the committee (pictured above) voted to decrease penalties for sexual 
harassment. 
Option 3: All male – increase (harassment_expand_allmale.jpg)
The Arusha City Council formed a special committee to consider legislation that would increase 
penalties for sexual harassment in the workplace. The law would require businesses to discipline 
employees found guilty of sexual harassment. Penalties could include immediate suspension or 
termination. 
The law would use a standard definition of sexual harassment, which includes unwelcome sexual 
advances, requests for sexual favors, and other verbal or physical harassment of a sexual nature. In 
Tanzania, most sexual harassment complaints are filed by women. 
Supporters claim the increase penalties will help end sexual harassment in the workplace. Those opposed 
to the law claim the law will excessively punish misunderstandings between employees. 
After careful discussion, the committee (pictured above) voted to increase penalties for sexual 
harassment. 
Option 4: All male – decrease (harassment_reduce_allmale.jpg)
The Arusha City Council formed a special committee to consider legislation that would decrease 
penalties for sexual harassment in the workplace. The law would restrict business’ ability to discipline 
employees found guilty of sexual harassment. Penalties could not include immediate suspension or 
termination of employment. 
The law would use a standard definition of sexual harassment, which includes unwelcome sexual 
advances, requests for sexual favors, and other verbal or physical harassment of a sexual nature. In 
Tanzania, most sexual harassment complaints are filed by women. 
Supporters claim the decreased penalties will protect employees who have experienced 
misunderstandings at work. Those opposed to the law claim the law will increase sexual harassment in 
the workplace. 
After careful discussion, the committee (pictured above) voted to decrease penalties for sexual 
harassment. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 34
QI_1 In the article about sexual harassment, of the eight members, how many committee 
members on the committee discussing the sexual harassment law were women? 
QI_2 I In the article about sexual harassment, what decision did the committee make 
regarding the sexual harassment law?
o Increasing penalties for sexual harassment (1) 
o Decreasing penalties for sexual harassment (2)
Q_F_1 Do you agree or disagree with the following statement: Personally, I think the 
committee made the right decision. 
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_F_2 Thinking about the gender composition of the committee, how fair was the 
decision-making process?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_F_3 Thinking about the gender composition of the committee, the committee can be 
trusted to make decisions that are right for the state’s citizens.
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q118
Please read the following hypothetical news story about laws regarding the number of 
special seats reserved for women in Parliament. You will be asked your opinion about the 
contents of the story. You will also be asked to recall some basic facts from the article. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 35
Option 1: Mixed panel – increase (quota_expand_mixedgender.jpg)
A special committee in the Tanzanian Parliament was formed to consider increasing the special seats 
quota for women members of parliament (MPs), in which parties appoint a number of women 
proportional to their vote share gained from elected constituencies. It would change the number of 
seats guaranteed to be held by women from 35% to 50%. 
Advocates of quota expansion claim it will result in more women MPs, counteracting voter and party 
officials’ bias against electing women candidates. More women MPs will facilitate gender equality, they 
argue, because women MPs do a better job representing women’s interests and serve a critical role 
model function that encourages women’s leadership throughout society. Those opposed to quota 
expansion hold that gender equality can best be achieved if women and men candidates compete in the 
same competitive electoral process. Many opponents argue that men and women politicians are equally 
qualified to represent citizens of both genders. Others add that the quota stigmatizes women politicians 
- fellow politicians and voters view them as unqualified, or perhaps mere agents of their party. 
The committee (pictured above) ultimately proposed to increase the women’s quota from 35% to 50%. 
Option 2: Mixed panel – decrease (quota_reduce_mixedgender.jpg)
A special committee in the Tanzanian Parliament was formed to consider decreasing the special seats 
quota for women members of parliament (MPs), in which parties appoint a number of women 
proportional to their vote share gained from elected constituencies. It would change the number of 
seats guaranteed to be held by women from 35% to 15%. 
Advocates of quota reduction hold that gender equality can best be achieved if women and men 
candidates compete in the same competitive electoral process. Many supporters argue that men and 
women politicians are equally qualified to represent citizens of both genders. Others add that the quota 
stigmatizes women politicians - fellow politicians and voters view them as unqualified, or perhaps mere
agents of their party. Those opposed to quota reduction claim it will result in fewer women MPs due to 
voter and party officials’ bias against electing women candidates. Fewer women MPs will undermine 
gender equality, they argue, because women MPs do a better job representing women’s interests and 
serve a critical role model function that encourages women’s leadership throughout society. 
The committee (pictured above) ultimately proposed to decrease the women’s quota from 35% to 15%. 
Option 3: All male – increase (quota_expand_allmale.jpg)
A special committee in the Tanzanian Parliament was formed to consider increasing the special seats 
quota for women members of parliament (MPs), in which parties appoint a number of women 
proportional to their vote share gained from elected constituencies. It would change the number of 
seats guaranteed to be held by women from 35% to 50%. 
Advocates of quota expansion claim it will result in more women MPs, counteracting voter and party 
officials’ bias against electing women candidates. More women MPs will facilitate gender equality, they 
argue, because women MPs do a better job representing women’s interests and serve a critical role 
model function that encourages women’s leadership throughout society. Those opposed to quota 
expansion hold that gender equality can best be achieved if women and men candidates compete in the 
same competitive electoral process. Many opponents argue that men and women politicians are equally 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 36
qualified to represent citizens of both genders. Others add that the quota stigmatizes women politicians 
- fellow politicians and voters view them as unqualified, or perhaps mere agents of their party. 
The committee (pictured above) ultimately proposed to increase the women’s quota from 35% to 50%. 
Option 4: All male – decrease (quota_reduce_allmale.jpg)
A special committee in the Tanzanian Parliament was formed to consider decreasing the special seats 
quota for women members of parliament (MPs), in which parties appoint a number of women 
proportional to their vote share gained from elected constituencies. It would change the number of 
seats guaranteed to be held by women from 35% to 15%. 
Advocates of quota reduction hold that gender equality can best be achieved if women and men 
candidates compete in the same competitive electoral process. Many supporters argue that men and 
women politicians are equally qualified to represent citizens of both genders. Others add that the quota 
stigmatizes women politicians - fellow politicians and voters view them as unqualified, or perhaps mere 
agents of their party. Those opposed to quota reduction claim it will result in fewer women MPs due to 
voter and party officials’ bias against electing women candidates. Fewer women MPs will undermine 
gender equality, they argue, because women MPs do a better job representing women’s interests and 
serve a critical role model function that encourages women’s leadership throughout society. 
The committee (pictured above) ultimately proposed to decrease the women’s quota from 35% to 15%. 
QI_1 In the article about special seats in Parliament, of the eight members, how many 
committee members on the committee discussing special seats policy were women? 
QI_2 I In the article about special seats in Parliament, what decision did the committee 
make regarding special seats for Parliament?
o Increasing the special seats reserved for women from 35% to 50% (1) 
o Decreasing the special seats reserved for women from 35% to 15% (2) 
Q_G_1 Do you agree or disagree with the following statement: Personally, I think the 
committee made the right decision. 
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 37
Q_G_2 Thinking about the gender composition of the committee, how fair was the 
decision-making process?
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
Q_G_3 Thinking about the gender composition of the committee, the committee can be 
trusted to make decisions that are right for the state’s citizens.
o Strongly disagree (1)
o Disagree (2) 
o Agree (3)
o Strongly Agree (4) 
o Refused (5)
transition_bias We will now ask you questions about discrimination and ethics in the 
journalism profession. 
Q_K_1 True or False: Unconscious bias always stems from the way we perceive differences 
in others and how we behave accordingly.
o True (1) 
o False (2) 
Q_K_2 What is affinity bias?
o Believing something because your peers believe it.
o Being more receptive of people who are like us in some way.
o Looking for information that backs up our beliefs about someone.
o Creating stereotypes based on groups people might belong to.
Q_K_3 What is confirmation bias?
o Believing something because your peers believe it.
o Looking for information that backs up our beliefs about someone.
o Creating stereotypes based on groups people might belong to.
o Being more receptive of people who are like us in some way.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 38
Q_K_4 True or False: Tanzania has legal protections ensuring gender equality in the 
media.
o True
o False
Q_K_5 Which of the following is true for interviewing youth? (select all that apply)
o Youth may provide consent to interview
o They often talk too much
o It is best to ask them Yes/No questions
o Interviewing youth requires additional patience
Q_K_6 Should you use marked or unmarked vehicles with radio station branding when 
conducting reporting and interviewing? 
o Always marked
o Always unmarked
o It depends on the situation
transition_scenario In this next section, we will ask you to comment on a set of hypothetical 
scenarios or pieces of media that are common in Tanzania. 
Randomized so either Q_J_male or Q_J_female appeared
Q_J_male Josephine is a student assigned to carry out a radio interview class project where 
they interview a local notable. He instead interviews a friend and pretends the friend is the 
notable. Should he be punished? What punishment would be most appropriate?
o No punishment (1) 
o A warning (2)
o A failing grade on the assignment (3) 
o A failing grade for the course (4) 
o Expulsion from the University (5) 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 39
Q_J_female Josephine is a student assigned to carry out a radio interview class project 
where they interview a local notable. She instead interviews a friend and pretends the 
friend is the notable. Should she be punished? What punishment would be most 
appropriate?
o No punishment (1) 
o A warning (2)
o A failing grade on the assignment (3) 
o A failing grade for the course (4) 
o Expulsion from the University (5) 
Randomized so that either Q_K_CCM, Q_K_Opp or Q_K_none appears, followed by 
Q_K_1 and Q_K_2
Q_K_CCM A reporter is covering a story on a CCM local councilor (Diwani) who has been 
accused of engaging in fraud and bribery. The CCM local councilor admits the bribery to 
the reporter but claims the story is more complicated. The CCM local councilor asks the 
reporter to delay the story until they can give their public account. When leaving the 
interview, the local councilor offers the reporter a brown envelope to cover the reporter’s 
expenses and time. 
Q_K_opp A reporter is covering a story on a ACT-Wazalendo local councilor (Diwani) who 
has been accused of engaging in fraud and bribery. The ACT-Wazalendo local councilor 
admits the bribery to the reporter but claims the story is more complicated. The ACT￾Wazalendo local councilor asks the reporter to delay the story until they can give their 
public account. When leaving the interview, the local councilor offers the reporter a 
brown envelope to cover the reporter’s expenses and time. 
Q_K_none A reporter is covering a story on local councilor (Diwani) who has been accused 
of engaging in fraud and bribery. The local councilor admits the bribery to the reporter but 
claims the story is more complicated. The local councilor asks the reporter to delay the 
story until they can give their public account. When leaving the interview, the local 
councilor offers the reporter a brown envelope to cover the reporter’s expenses and time. 
Q_K_1 What should the reporter do with the brown envelope?
o Accept (1)
o Decline (2) 
Q_K_2 What should the reporter do with the story? 
o Run the story immediately (1) 
o Run the story after the diwani gives a public account (2) 
o Do not run the story (3)
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 40
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
Randomized so that either Q_L_CCM, Q_L_Opp or Q_L_none appears, followed by Q_L_1 
Q_L_CCM A reporter is covering a story on a CCM Parliamentarian who rarely shows up 
to sessions and seldom speaks on the floor. The CCM Parliamentarian tells the reporter to 
stop the story and that if the reporter does not, the MP will get them fired and will send 
the police after them. 
Q_L_opp A reporter is covering a story on a CHADEMA Parliamentarian who rarely 
shows up to sessions and seldom speaks on the floor. The CHADEMA Parliamentarian tells 
the reporter to stop the story and that if the reporter does not, the MP will get them fired 
and will send the police after them. 
Q_L_none A reporter is covering a story on a Parliamentarian who rarely shows up to 
sessions and seldom speaks on the floor. The Parliamentarian tells the reporter to stop 
the story and that if the reporter does not, the MP will get them fired and will send the 
police after them. 
Q_L_1 What should the reporter do? 
o Definitely run the story (1)
o Probably run the story (2)
o Probably not run the story (3)
o Definitely not run the story (4)
Randomized so either Q_M_male or Q_M_female appeared
Q_M_male A reporter is covering a story about Patrick Mpiri—a famous businessperson 
who is rumored to be carrying on affair. When they are interviewed by the reporter, 
Patrick tells the reporter that the rumors are true. Patrick asks the reporter to not run 
the story because of the embarrassment the story will cause him toward family and 
constituents. What should the reporter do? 
o Definitely run the story (1)
o Probably run the story (2)
o Probably not run the story (3)
o Definitely not run the story (4)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 41
Q_M_female A reporter is covering a story about Catherine Mpiri—a famous 
businessperson who is rumored to be carrying on affair. When they are interviewed by the 
reporter, Catherine tells the reporter that the rumors are true. Catherine asks the 
reporter to not run the story because of the embarrassment the story will her toward 
family and constituents. What should the reporter do? 
o Definitely run the story (1) 
o Probably run the story (2) 
o Probably not run the story (3) 
o Definitely not run the story (4) 
Q_N_1 How confident are you that you could independently, without any input from 
others, produce a 5 minute radio spot? 
o I am completely able (1) 
o I am probably able (2) 
o I am probably unable (3) 
o I am completely unable (4) 
Q_N_2 Please indicate the extent to which you agree or disagree with the following 
statements.
(1)
Strongly 
agree
(2)
Agree
(3)
Disagree
(4)
Strongly 
Disagree
(5)
Refused
I am well qualified to conduct investigative 
journalism. 
I have a good understanding of the 
important political issues facing our country.
I have a good understanding of the 
important gender and youth issues facing 
our country.
I could do as good a job as a journalist as 
most other people.
People like me don’t have any say about 
what government does
Public officials don’t care much about what 
journalists like me think
Other students look to me as a leader.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 42
(1)
Strongly 
agree
(2)
Agree
(3)
Disagree
(4)
Strongly 
Disagree
(5)
Refused
I am motivated to mobilize my community 
to take action about important matters.
One day in the future, I might consider 
running for political office. 
I can identify fake news easily.
Q_N_3 Please indicate your strength in performing the following:
Area of expertise (4)
Strong
(3)
Average
(2)
Poor
(1)
Very 
Poor
(99)
Decline to 
Answer
Radio journalism
Gender sensitive reporting
Finding the story
Covering the story
Making the story
Q_O_1 Have you ever considered running for office in student government?
o Not at all (1)
o Rarely (2)
o Sometimes (3) 
o Frequently (4) 
o Refused (5) 
Q_O_2 Have you ever considered running for political office?
o Not at all (1)
o Rarely (2)
o Sometimes (3) 
o Frequently (4) 
o Refused (5) 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 43
Q_O_3 How often in the last month have you had discussions with others about sexual 
harassment?
o Not at all (1)
o Rarely (2)
o Sometimes (3) 
o Frequently (4) 
o Refused (5) 
Q_O_4 How often in the last month have you had discussions with others about political 
gender quotas?
o Not at all (1)
o Rarely (2)
o Sometimes (3) 
o Frequently (4) 
o Refused (5) 
Q_O_5 How often in the last month have you had discussions with others about fake news?
o Not at all (1)
o Rarely (2)
o Sometimes (3) 
o Frequently (4) 
o Refused (5) 
Q_O_6 When you’re with other people, how often do you put on a show to impress or 
entertain them?
o Always
o Most of the time
o Half of the time 
o Once in a while 
o Never 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 44
Q_O_7 How good or bad of an actor would you be?
o Excellent 
o Good
o Fair
o Poor
o Very Poor
Q_O_8 When you’re in a group of people, how often are you the center of attention?
o Always
o Most of the time
o Half of the time 
o Once in a while 
o Never 
QO_9 Below is a list of the various political parties in Tanzania. For each party, please 
indicate how close you feel.
Area of expertise Not at all 
close
A little 
close Average Very 
close
Extremely 
close
CCM
CUF
CHADEMA
ACT-Wazalendo
Other Party
Q_O_10 What is your constituency? 
Q_O_11 What is your local council? 
Q_O_12 What is the name of your current Member of Parliament? 
Q_O_13 What is the name of your current local councilor? 
QX_0 Please confirm which of these is true for you
o I was in the journalism training program (This Year) (1)
o I was not in the journalism training program (This Year) (2)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 45
QX_1 Did you have a training program/internship between June and October? 
o Yes, I had a training program/ internship (1)
o No, I didn't have a training program / internship (2)
QX_2 Was the training program / internship related to journalism studies (such as 
journalism, media, and public relations) or a completely different area?
o Same area (1)
o Related Area (2)
o Unrelated (3)
o Not applicable (4)
o I refuse to answer (5) 
QX_3 What organization did you do your training program/internship with? 
QX_4 Were you employed between June and October? 
o Yes, I was employed (1) 
o No, I was not employed (2) 
QX_5 Was your job related to journalism studies (such as journalism, media, and public 
relations) or a completely different area?
o Same area (1)
o Related Area (2)
o Unrelated (3)
o Not applicable (4)
o I refuse to answer (5)
QX_6 What organization did you work for? 
QX_7 What position did you have in this organization? 
QX_8 Did you complete the questionnaire from "NORC at the University of Chicago" in 
October 2019?
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 46
QX_9 Why didn't you complete the questionnaire? Please select all that apply.
o I didn’t receive an email about the questionnaire
o I didn’t get an SMS about the questionnaire 
o The participation compensation wasn’t enough money
o I didn’t have a computer to complete the questionnaire
o I didn’t have network to complete the questionnaire
o I wanted to do it but forgot
o Other (provide response)
QZ_2 Did you discuss the journalism training provided by Internews with students who 
were selected for the training?
o Never (1)
o Rarely (2) 
o Several Times (3)
o Frequently (4) 
QZ_3 What did you learn from them?
QY_1A The name of your trainer (fill in)
QY_1B How would you rate your instructor’s background knowledge in radio journalism in 
teaching the course?
o Not at all knowledgeable (1)
o Not very knowledgeable (2) 
o Somewhat knowledgeable (3) 
o Very knowledgeable (4) 
o Extremely knowledgeable (5) 
QY_1C How would you rate your instructor’s background knowledge in “diversity and 
inclusion” in Tanzanian society in teaching the course? 
o Not at all knowledgeable (1)
o Not very knowledgeable (2) 
o Somewhat knowledgeable (3) 
o Very knowledgeable (4) 
o Extremely knowledgeable (5) 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 47
QY_1D How would you rate your instructor’s typical teaching across the following 
characteristics? Please evaluate on a scale from 1-5 where 5 is extremely and 1 is not at all.
Rating Very bad Bad Average Good
Very 
Good Refused
Enthusiasm regarding 
the material
Effectiveness in 
communicating ideas
Receptiveness in 
answering student 
questions 
Equitable treatment of 
male and female 
students:
Encourage students to 
speak in class:
QY_2A Would you rate the pace of class lessons too slow, too fast, or about right?
o The pace was much too fast (1)
o The pace was a little fast (2)
o The pace was about right (3) 
o The pace was a little slow (4)
o The pace was much too slow (5)
QZ_1 Have you done any of the following activities during the break period? Select all that 
apply
o Creating a radio segment with others
o Creating a radio segment by yourself
o Received any practical training on making radio segment
o Received any practical training on gender and youth issues in Tanzania
o Helped your family with farming, a home business, and household work
o Entertainment 
o Other
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 48
QZ_4 What topics are you most interested in reporting about? Please select up to three 
responses. 
o Rural issues (1) 
o Youth (2) 
o Women (3) 
o Sports (4) 
o Politics (5) 
o Economy (6) 
o Entertainment (7)
o Other (8)
QZ_5 Have you changed the email or phone number you use in the last six months? If 
YES write them down below . If not, write "NO" 
o Email / Email [fill] (5) ________________________________________________
o Telephone number [fill] (6) ___________________________________________
QZ_6 Did you know you can rent a recorder from college for recording interviews?
o Yes (1)
o No (2)
QZ_7 Did you know that the college has computers with Adobe Audition that you can 
reserve to use?
o Yes (1)
o No (2) 
QZ_8 Do you know how to use Adobe Audition software to edit a radio audio segment?
o Yes (1)
o No (2)
SJMC students can rent a recorder from the campus throughout the day for recording interviews. There 
are 20 computers with Adobe Audition that you can reserve. To use these recorders or computers, 
contact Mr. Fanuel Hume of IT. His phone number is 0713549992. 
SAU-T students can rent a recorder from the campus for recording interviews; the time will depend on 
the project. There are 50 computers with Adobe Audition that you can reserve. To use these recorders 
or computers, contact Dotto Bulendu. His phone numbers are 0787373550 or 0629093725. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 49
QZ_10 Would you like and be able to participate in a contest to create a 5-minute audio 
segment at the end of the year, in which the top ten winners receive prizes and hose who 
have the highest-quality segments will be broadcast on the BBC? 
o Yes (1)
o No (2)
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 50
IAT COMPONENT
WHAT IS "IAT"? 
The implicit association test (IAT) measures the strength of associations between concepts 
(e.g., fat people, old people) and evaluations (e.g., good, bad) or stereotypes (e.g., athletic, 
clumsy). The main idea is that making a response is easier when closely related items share the 
same response key. 
When doing an IAT you are asked to quickly sort words into categories that are on the left and 
right hand side of the computer screen by pressing the “E" key if the word belongs to the 
category on the left and the “I” key if the word belongs to the category on the right. The IAT 
has five main parts. 
In the first part of the IAT, you sort words relating to the concepts (e.g., fat, thin) into 
categories. So if the category “Thin” was on the left, and a word related to being thin appeared 
on the screen, you would press the “E” key. 
In the second part of the IAT, you sort words relating to the evaluation (e.g., good, bad). So if 
the category “Good” was on the left, and a pleasant word appeared on the screen, you would 
press the “E” key. 
In the third part of the IAT, the categories are combined and you are asked to sort both 
concept and evaluation words. So the categories on the left hand side would be Thin/Good and 
the categories on the right hand side would be Fat/Bad. 
In the fourth part of the IAT, the placement of the concepts switches. If the category “Thin” 
was previously on the left, now it would be on the right. 
In the final part of the IAT, the categories are combined in a way that is opposite what they 
were before. If the category on the left was previously Thin/Good, it would now be Thin/Bad. 
The IAT score is based on how long it takes a person, on average, to sort the words in the 
third part of the IAT versus the fifth part of the IAT. We would say that one has an implicit 
preference for thin people relative to fat people if they are faster to categorize words when 
Thin and Good share a response key and Fat and Bad share a response key, relative to the 
reverse. 
The information you provide is confidential. Only the researchers from NORC will have access 
to the data. The information you and others provide will be summarized in a report. Your 
name or any other identifying information will never be used when we report results of the 
test. 
Your participation in this test is completely voluntary – it is your choice to take this test or not. 
If you feel uneasy with any of the questions, you may stop the test at any time. If you decide not 
to take part or to stop the test, you will not lose any services that you are otherwise receiving. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 51
By clicking the ‘Take the test’ button at the bottom right of the screen, you 
acknowledge your acceptance to participate in this test. Thank you very much for 
your participation. 
Let's start with a warm-up IAT. In this study you will complete an Implicit Association Test 
(IAT) in which you will be asked to sort words into groups as fast as you can. This study should 
take about 15 minutes to complete. 
1. Instructions: Place your left and right index fingers on the E and I keys. At the top of the 
screen are 2 categories. In the task, words and/or images appear in the middle of the 
screen. 
When the word/image belongs to the category on the left, press the E key as fast as you 
can. When it belongs to the category on the right, press the I key as fast as you can. If you 
make an error, a red X will appear. Correct errors by hitting the other key. 
Please try to go as fast as you can while making as few errors as possible. 
When you are ready, please press the [Space] bar to begin. 
2. Now, the categories have changed, but the rules remain the same. Please try to go as fast as 
you can while making as few errors as possible. Correct errors by hitting the other key. 
When you are ready, please press the [Space] bar to begin. 
3. Now the four categories you saw separately will appear together. Remember, each 
word/image fits in only one of the four categories. The label/item colors may help you 
identify the appropriate category. 
Use the E key for the two categories on the left and the I key for the two categories on the 
right. Again, try to go as fast as possible without making mistakes. Correct errors by hitting 
the other key. Practice this combination now. 
When you are ready, please press the [Space] bar to begin. 
4. Please continue the task as you were just doing it. Again, try to go as fast as possible 
without making mistakes. Correct errors by hitting the other key. 
When you are ready, please press the [Space] bar to begin. 
5. Notice the categories from before have switched sides. Please practice this new 
configuration now. Remember to try to go as fast as you can while making as few errors as 
possible. Correct errors by hitting the other key. 
When you are ready, please press the [Space] bar to begin. 
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 52
6. Notice the four categories have been combined again, but in a new configuration. Please 
practice this combination now, and remember to go as fast as you can while making as few 
mistakes as possible.Correct errors by hitting the other key. 
When you are ready, please press the [Space] bar to begin. 
7. Please continue the task as you were just doing it, and remember to go as fast as you can 
while making as few mistakes as possible. Correct errors by hitting the other key. 
When you are ready, please press the [Space] bar to begin. 
Press E or I to advance to the next word/image. Correct mistakes by pressing the other key. 
Good job! Now we can begin with the real IAT. Again, you will be asked to sort words into 
groups as fast as you can. 
That marks the end of the test. Thank you again for your participation.
CONTRACT NO. GS-10F-0033M / ORDER NO. AID-OAA-M-13-00013 / DRG-LER I TASKING N061
USAID.GOV Impact Evaluation of an Intensive Journalism Training Activity In Tanzania: Final Report | 53
APPENDIX B: REGRESSION RESULTS
1 Knowledge of Uni Resource Access 
(1) (2) (3) 
Access Uni Recorders? Access Uni Computers? Use Edit Software? 
b/se b/se b/se 
main 
Program -0.07 0.73 2.19*** 
(0.233) (0.771) (0.638) 
Constant 1.20*** 3.72*** 2.26*** 
(0.183) (0.507) (0.263) 
Observations 428 428 428 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 1: Knowledge of Uni Resource Access Logistic Regression Results 
No
Yes
Predicted Probabilities
Control
Program
Treatment
Treatment Effects uniqz_6
No
Yes
Predicted Probabilities
Control
Program
Treatment
Treatment Effects uniqz_7
No
Yes
Predicted Probabilities
Control
Program
Treatment
Treatment Effects uniqz_8
Figure 1: Knowledge of Uni Resource Access Predicted Values 
2 Question 1: Content Battery 
4 
(1) 
Q1.1 Correct 
(2) 
Q1.2 Correct 
(3) 
Q1.3 Correct 
(4) 
Q1.4 Correct 
main 
b/se b/se b/se b/se 
Program 
Constant 
0.18 
(0.212) 
1.10*** 
0.42** 
(0.187) 
0.35** 
-0.57 
(0.420) 
-2.71*** 
-0.41 
(0.366) 
-2.48*** 
(0.160) (0.141) (0.287) (0.260) 
Observations 512 512 512 512 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 2: Q1 Battery Logistic Regression Results 
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ1_1
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ1_2
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ1_3
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ1_4_total Figure 2: Q1 Battery Predicted Probabilities 
5 
3 Question 2: Production Battery 
(1) (2) 
corrQ2 1 corrQ2 2 total 
b/se b/se 
main 
Program 0.30 0.32 
(0.194) (0.240) 
Constant -0.88*** -1.70*** 
(0.152) (0.192) 
Observations 512 512 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 3: Q2 Battery Logistic Regression Results 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ2_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQ2_2_total
Figure 3: Q2 Battery Predicted Probabilities 
6 
4 Question K: Ethics Battery 
(1) 
QK.1 Correct 
(2) 
QK.2 Correct 
(3) 
QK.3 Correct 
(4) 
QK.4 Correct 
(5) 
QK.5 Correct 
main 
b/se b/se b/se b/se b/se 
Program 
Constant 
-0.53 
(0.347) 
-2.30*** 
0.41* 
(0.211) 
-1.29*** 
0.11 
(0.200) 
-0.97*** 
0.07 
(0.190) 
-0.68*** 
-0.17 
(0.263) 
-1.78*** 
(0.241) (0.169) (0.156) (0.147) (0.198) 
Observations 512 512 512 512 512 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 4: QK Battery Logistic Regression Results 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQK_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQK_2
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQK_3
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Treatment Effects corrQK_4
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Treatment Effects corrQK_5_total
Figure 4: QK Battery Predicted Probabilities 
7 
5 Implicit Association Test 
(1) 
1 Dscore 
b/se 
Program 
Constant 
0.05 
(0.033) 
0.15*** 
(0.027) 
Observations 391 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 5: IAT Regression Results 
-1.5-1.25-1 -.75 -.5 -.25
0 .25 .5 .75 1 1.251.5
Predicted Values
Control
Program
Treatment
Treatment Effects IAT
Figure 5: IAT Predicted Values 
8 
6 Question N: Effcacy Battery 
(1) 
Radio Journalism 
b/se 
(2) 
Gender Sensitive Reporting 
b/se 
(3) 
Finding Story 
b/se 
(4) 
Covering Story 
b/se 
(5) 
Making Story 
b/se 
Program 
Constant 
0.08* 
(0.040) 
2.70*** 
0.06 
(0.045) 
2.62*** 
0.03 
(0.039) 
2.75*** 
0.05 
(0.041) 
2.71*** 
0.15*** 
(0.044) 
2.61*** 
(0.032) (0.035) (0.030) (0.032) (0.036) 
Observations 511 511 511 511 510 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 6: Journalism Ability Self Reported Strength Regression Results 
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn3_1
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn3_2
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn3_3
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn3_4
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn3_5
Figure 6: Journalism Ability Self Reported Strength Predicted Values 
9 
(1) 
Qualifed Investigative Journalism 
b/se 
(2) 
My Skill v. Most 
b/se 
(3) 
Can Mobilize Community 
b/se 
(4) 
Can Identify Fake News 
b/se 
Program 
Constant 
-0.07 
(0.065) 
2.31*** 
-0.02 
(0.046) 
2.71*** 
0.05 
(0.041) 
2.71*** 
0.14*** 
(0.050) 
2.58*** 
(0.050) (0.034) (0.032) (0.042) 
Observations 498 510 512 511 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 7: Journalism Effcacy Regression Results 
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn2_1
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn2_4
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn2_9
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Treatment Effects selfqn2_11
Figure 7: Journalism Effcacy Predicted Values 
10 
7 Topical Interest for Journalism 
(1) 
Rural 
b/se 
(2) 
Youth 
b/se 
(3) 
Women 
b/se 
(4) 
Sports 
b/se 
(5) 
Politics 
b/se 
(6) 
Economy 
b/se 
(7) 
Entertainment 
b/se 
main 
Program 
Constant 
0.59*** 
(0.200) 
-0.11 
0.21 
(0.220) 
0.84*** 
0.53*** 
(0.203) 
0.13 
-0.42* 
(0.235) 
-1.01*** 
-0.21 
(0.263) 
-1.49*** 
-0.30 
(0.338) 
-2.13*** 
-0.66*** 
(0.236) 
-0.90*** 
(0.154) (0.168) (0.154) (0.174) (0.199) (0.250) (0.170) 
Observations 428 428 428 428 428 428 428 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 8: Topical Interest Logistic Regression Results 
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_1
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_2
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_3
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_4
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_5
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_6
No
Yes
Control Predicted Probabilities
Program
Treatment
Treatment Effects topicqz4_7
Figure 8: Topical Interest Logistic Predicted Values 
11 
8 Willingness to Participate in Contest to Produce Short Radio Seg￾ments 
(1) 
Participate Contest 
b/se 
Program 0.00 
(0.025) 
Constant 1.96*** 
(0.019) 
Observations 428 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 9: Participate in Contest Regression Results 
0 1 2
Predicted Values
Control
Program
Treatment
Treatment Effects Participate Contest
Figure 9: Participate in Contest Predicted Values 
12 
9 Other Statistical Analysis 
9.1 Balance Checks Program and No Program 
9.2 Heterogeneous Treatment Effects: Gender 
9.2.1 Question 1: Content Battery 
(1) 
Q1.1 Correct 
(2) 
Q1.2 Correct 
(3) 
Q1.3 Correct 
(4) 
Q1.4 Correct 
b/se b/se b/se b/se 
main 
Program 
Female 
0.22 
(0.279) 
-0.15 
0.32 
(0.244) 
-0.27 
-0.58 
(0.618) 
0.67 
-0.40 
(0.515) 
0.37 
Program × Female 
Constant 
(0.331) 
-0.05 
(0.434) 
1.15*** 
(0.293) 
0.19 
(0.386) 
0.47** 
(0.577) 
-0.25 
(0.865) 
-2.99*** 
(0.540) 
0.06 
(0.746) 
-2.68*** 
(0.210) (0.184) (0.419) (0.366) 
Observations 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 10: Q1 Battery Logistic Regression Results 
13 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ1_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ1_2
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ1_3
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ1_4_total
Figure 10: Q1 Battery Predicted Probabilities 
14 
9.2.2 Question 2: Production Battery 
main 
Program 
Female 
Program × Female 
Constant 
Observations 
(1) 
Q2.1 Correct 
b/se 
0.34 
(0.244) 
-0.38 
(0.326) 
-0.08 
(0.411) 
-0.75*** 
(0.192) 
508 
(2) 
Q2.2 Correct 
b/se 
-0.10 
(0.304) 
-0.68 
(0.439) 
1.10** 
(0.526) 
-1.49*** 
(0.231) 
508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 11: Q2 Battery Logistic Regression Results 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ2_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQ2_2_total
Figure 11: Q2 Battery Predicted Probabilities 
15 
9.2.3 Question K: Ethics Battery 
(1) 
QK.1 Correct 
(2) 
QK.2 Correct 
(3) 
QK.3 Correct 
(4) 
QK.4 Correct 
(5) 
QK.5 Correct 
b/se b/se b/se b/se b/se 
main 
Program 
Female 
-0.32 
(0.435) 
0.17 
0.30 
(0.268) 
-0.24 
0.31 
(0.265) 
0.46 
-0.06 
(0.235) 
-1.05*** 
-0.20 
(0.320) 
-0.44 
Program × Female 
Constant 
(0.489) 
-0.67 
(0.733) 
-2.34*** 
(0.357) 
0.26 
(0.440) 
-1.20*** 
(0.319) 
-0.53 
(0.410) 
-1.15*** 
(0.334) 
0.46 
(0.418) 
-0.31* 
(0.428) 
0.02 
(0.566) 
-1.60*** 
(0.316) (0.212) (0.210) (0.181) (0.239) 
Observations 508 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 12: QK Battery Logistic Regression Results 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQK_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQK_2
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQK_3
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQK_4
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Male Female
Treatment Effects corrQK_5_total
Figure 12: QK Battery Predicted Probabilities 
16 
9.2.4 Implicit Association Test 
Program 
Female 
Program × Female 
Constant 
Observations 
(1) 
1 Dscore 
b/se 
0.07* 
(0.039) 
0.21*** 
(0.047) 
-0.05 
(0.059) 
0.07** 
(0.030) 
468 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 13: IAT Regression Results 
-1.5-1.25-1 -.75-.5-.25
0 .25 .5 .75 1 1.251.5
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects IAT
Figure 13: IAT Predicted Values 
17 
9.2.5 Question N: Effcacy Battery 
(1) 
Radio Journalism 
b/se 
(2) 
Gender Sensitive Reporting 
b/se 
(3) 
Finding Story 
b/se 
(4) 
Covering Story 
b/se 
(5) 
Making Story 
b/se 
Program 
Female 
0.10** 
(0.051) 
-0.06 
0.03 
(0.060) 
0.03 
0.09* 
(0.051) 
0.04 
0.04 
(0.053) 
-0.03 
0.11** 
(0.054) 
-0.16** 
Program × Female 
Constant 
(0.068) 
-0.04 
(0.085) 
2.71*** 
(0.072) 
0.09 
(0.090) 
2.60*** 
(0.063) 
-0.12 
(0.080) 
2.73*** 
(0.069) 
0.03 
(0.086) 
2.71*** 
(0.077) 
0.13 
(0.093) 
2.67*** 
(0.041) (0.047) (0.040) (0.041) (0.044) 
Observations 507 507 507 507 506 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 14: Journalism Ability Self Reported Strength Regression Results 
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn3_1
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn3_2
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn3_3
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn3_4
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn3_5
Figure 14: Journalism Ability Self Reported Strength Predicted Values 
18 
(1) 
Qualifed Investigative Journalism 
b/se 
(2) 
My Skill v. Most 
b/se 
(3) 
Can Mobilize Community 
b/se 
(4) 
Can Identify Fake News 
b/se 
Program 
Female 
-0.07 
(0.082) 
-0.19* 
-0.05 
(0.057) 
-0.09 
0.03 
(0.051) 
-0.10 
0.13** 
(0.061) 
-0.16* 
Program × Female 
Constant 
(0.106) 
0.03 
(0.136) 
2.38*** 
(0.075) 
0.07 
(0.096) 
2.75*** 
(0.068) 
0.06 
(0.086) 
2.74*** 
(0.089) 
0.07 
(0.105) 
2.64*** 
(0.064) (0.040) (0.041) (0.051) 
Observations 494 506 508 507 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 15: Journalism Effcacy Regression Results 
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn2_1
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn2_4
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn2_9
Strongly Disagree
Disagree
Agree
Strongly Agree
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects selfqn2_11
Figure 15: Journalism Effcacy Predicted Values 
19 
9.2.6 Topical Interest for Journalism 
(1) 
Rural 
b/se 
(2) 
Youth 
b/se 
(3) 
Women 
b/se 
(4) 
Sports 
b/se 
(5) 
Politics 
b/se 
(6) 
Economy 
b/se 
(7) 
Entertainment 
b/se 
main 
Program 
Female 
0.55** 
(0.235) 
-0.17 
0.37 
(0.256) 
0.28 
0.85*** 
(0.241) 
2.29*** 
-0.47* 
(0.257) 
-1.49*** 
-0.27 
(0.266) 
-1.54*** 
-0.76* 
(0.416) 
0.19 
-0.72*** 
(0.278) 
-0.13 
Program × Female 
Constant 
(0.291) 
0.19 
(0.377) 
-0.14 
(0.319) 
-0.30 
(0.417) 
0.72*** 
(0.358) 
-0.56 
(0.472) 
-0.68*** 
(0.419) 
0.82 
(0.523) 
-0.68*** 
(0.470) 
-0.21 
(0.654) 
-0.94*** 
(0.427) 
0.35 
(0.609) 
-1.99*** 
(0.322) 
0.26 
(0.443) 
-0.87*** 
(0.180) (0.191) (0.190) (0.190) (0.199) (0.276) (0.196) 
Observations 508 508 508 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 16: Topical Interest Logistic Regression Results 
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_1
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_2
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_3
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_4
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_5
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_6
No
Yes
Predicted Probabilities Control
Program
Treatment
Male Female
Treatment Effects topicqz4_7
Figure 16: Topical Interest Logistic Predicted Values 
20 
9.2.7 Willingness to Participate in Contest to Produce Short Radio Segments 
(1) 
Participate Contest 
b/se 
Program 0.00 
(0.025) 
Female -0.06 
(0.049) 
Program × Female 0.03 
(0.060) 
Constant 1.97*** 
(0.019) 
Observations 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 17: Participate in Contest Regression Results 
0 1 2
Predicted Values
Control
Program
Treatment
Male Female
Treatment Effects Participate Contest
Figure 17: Participate in Contest Predicted Values 
9.3 Heterogeneous Treatment Effects: Major 
9.3.1 Question 1: Content Battery 
21 
(1) 
Q1.1 Correct 
(2) 
Q1.2 Correct 
(3) 
Q1.3 Correct 
(4) 
Q1.4 Correct 
main 
b/se b/se b/se b/se 
Program 
MassComm 
-0.26 
(0.301) 
-0.07 
0.45* 
(0.264) 
0.13 
-0.78 
(0.572) 
-1.09 
-0.48 
(0.535) 
0.37 
PR 
(0.454) 
-0.47 
(0.389) 
0.02 
(1.088) 
0.01 
(0.658) 
-0.23 
Program × MassComm 
Program × PR 
Constant 
(0.361) 
1.29* 
(0.676) 
0.92* 
(0.492) 
1.27*** 
(0.322) 
0.39 
(0.542) 
-0.40 
(0.428) 
0.34 
(0.609) 
0.49 
(1.542) 
0.07 
(0.944) 
-2.54*** 
(0.648) 
-0.14 
(0.959) 
0.55 
(0.875) 
-2.54*** 
(0.247) (0.207) (0.393) (0.393) 
Observations 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 18: Q1 Battery Logistic Regression Results 
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ1_1
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ1_2
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ1_3
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ1_4_total
Figure 18: Q1 Battery Predicted Probabilities 
22 
9.3.2 Question 2: Production Battery 
main 
Program 
MassComm 
PR 
Program × MassComm 
Program × PR 
Constant 
Observations 
(1) 
Q2.1 Correct 
b/se 
0.12 
(0.288) 
0.35 
(0.412) 
0.24 
(0.351) 
0.58 
(0.525) 
0.39 
(0.451) 
-1.04*** 
(0.233) 
508 
(2) 
Q2.2 Correct 
b/se 
0.32 
(0.348) 
0.25 
(0.508) 
0.01 
(0.449) 
0.29 
(0.624) 
-0.30 
(0.581) 
-1.77*** 
(0.289) 
508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 19: Q2 Battery Logistic Regression Results 
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ2_1
0 .2 .4 .6 .8 1
Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQ2_2_total
Figure 19: Q2 Battery Predicted Probabilities 
23 
9.3.3 Question K: Ethics Battery 
(1) 
QK.1 Correct 
(2) 
QK.2 Correct 
(3) 
QK.3 Correct 
(4) 
QK.4 Correct 
(5) 
QK.5 Correct 
b/se b/se b/se b/se b/se 
main 
Program 
MassComm 
-0.64 
(0.447) 
-0.87 
0.46 
(0.320) 
0.60 
-0.11 
(0.277) 
-0.42 
0.20 
(0.274) 
0.26 
0.03 
(0.371) 
0.65 
PR 
(0.794) 
-0.29 
(0.446) 
0.36 
(0.440) 
-0.31 
(0.401) 
0.36 
(0.484) 
-0.16 
Program × MassComm 
Program × PR 
Constant 
(0.535) 
0.00 
(.) 
0.47 
(0.751) 
-2.04*** 
(0.392) 
0.16 
(0.555) 
-0.19 
(0.495) 
-1.54*** 
(0.357) 
1.16** 
(0.547) 
-0.18 
(0.483) 
-0.79*** 
(0.335) 
0.30 
(0.514) 
-0.57 
(0.444) 
-0.84*** 
(0.481) 
-1.62** 
(0.798) 
0.15 
(0.621) 
-1.85*** 
(0.321) (0.268) (0.220) (0.222) (0.299) 
Observations 456 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 20: QK Battery Logistic Regression Results 
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQK_1
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQK_2
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQK_3
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQK_4
0 .2 .4 .6 .8 1 Predicted Probability
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects corrQK_5_total
Figure 20: QK Battery Predicted Probabilities 
24 
9.3.4 Implicit Association Test 
Program 
MassComm 
PR 
Program × MassComm 
Program × PR 
Constant 
Observations 
(1) 
1 Dscore 
b/se 
0.11** 
(0.044) 
0.11* 
(0.066) 
0.07 
(0.054) 
-0.14* 
(0.082) 
-0.10 
(0.070) 
0.11*** 
(0.036) 
468 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 21: IAT Regression Results 
-1.5-1.25-1-.75-.5-.25
0 .25 .5 .75 11.251.5
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects IAT
Figure 21: IAT Predicted Values 
25 
9.3.5 Question N: Effcacy Battery 
(1) 
Radio Journalism 
b/se 
(2) 
Gender Sensitive Reporting 
b/se 
(3) 
Finding Story 
b/se 
(4) 
Covering Story 
b/se 
(5) 
Making Story 
b/se 
Program 
MassComm 
0.09* 
(0.056) 
0.12 
-0.01 
(0.062) 
-0.09 
0.03 
(0.055) 
0.02 
0.09 
(0.057) 
0.11 
0.10* 
(0.059) 
-0.05 
PR 
(0.078) 
-0.10 
(0.094) 
-0.15* 
(0.081) 
-0.03 
(0.081) 
-0.03 
(0.099) 
-0.22*** 
Program × MassComm 
Program × PR 
Constant 
(0.077) 
-0.03 
(0.095) 
-0.02 
(0.100) 
2.70*** 
(0.082) 
0.15 
(0.118) 
0.15 
(0.106) 
2.68*** 
(0.071) 
0.05 
(0.101) 
0.00 
(0.094) 
2.75*** 
(0.078) 
-0.10 
(0.104) 
-0.07 
(0.101) 
2.69*** 
(0.082) 
0.04 
(0.119) 
0.13 
(0.104) 
2.69*** 
(0.047) (0.050) (0.044) (0.048) (0.050) 
Observations 507 507 507 507 506 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 22: Journalism Ability Self Reported Strength Regression Results 
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn3_1
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn3_2
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn3_3
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn3_4
Very Poor
Poor
Average
Strong
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn3_5
Figure 22: Journalism Ability Self Reported Strength Predicted Values 
26 
(1) 
Qualifed Investigative Journalism 
b/se 
(2) 
My Skill v. Most 
b/se 
(3) 
Can Mobilize Community 
b/se 
(4) 
Can Identify Fake News 
b/se 
Program 
MassComm 
-0.12 
(0.091) 
-0.25* 
-0.00 
(0.056) 
-0.00 
0.03 
(0.059) 
-0.00 
0.11 
(0.070) 
0.04 
PR 
(0.149) 
0.01 
(0.105) 
-0.09 
(0.087) 
-0.04 
(0.105) 
-0.10 
Program × MassComm 
Program × PR 
Constant 
(0.112) 
0.18 
(0.192) 
0.08 
(0.145) 
2.35*** 
(0.076) 
-0.09 
(0.138) 
-0.05 
(0.107) 
2.75*** 
(0.075) 
0.02 
(0.114) 
0.07 
(0.094) 
2.72*** 
(0.100) 
-0.04 
(0.127) 
0.14 
(0.118) 
2.60*** 
(0.072) (0.045) (0.048) (0.060) 
Observations 494 506 508 507 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 23: Journalism Effcacy Regression Results 
Strongly Disagree
Disagree
Agree
Strongly Agree Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn2_1
Strongly Disagree
Disagree
Agree
Strongly Agree Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn2_4
Strongly Disagree
Disagree
Agree
Strongly Agree Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn2_9
Strongly Disagree
Disagree
Agree
Strongly Agree Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects selfqn2_11
Figure 23: Journalism Effcacy Predicted Values 
27 
9.3.6 Topical Interest for Journalism 
(1) 
Rural 
b/se 
(2) 
Youth 
b/se 
(3) 
Women 
b/se 
(4) 
Sports 
b/se 
(5) 
Politics 
b/se 
(6) 
Economy 
b/se 
(7) 
Entertainment 
b/se 
main 
Program 
MassComm 
0.26 
(0.255) 
-0.01 
0.04 
(0.295) 
-0.52 
0.55** 
(0.263) 
-0.05 
0.06 
(0.318) 
0.04 
-0.06 
(0.364) 
0.56 
-0.11 
(0.469) 
0.88 
-0.68** 
(0.288) 
-0.24 
PR 
(0.380) 
-0.50 
(0.409) 
-0.49 
(0.382) 
-0.27 
(0.473) 
0.73** 
(0.478) 
0.75* 
(0.558) 
0.75 
(0.417) 
-0.57 
Program × MassComm 
Program × PR 
Constant 
(0.324) 
0.18 
(0.498) 
1.19*** 
(0.436) 
-0.04 
(0.347) 
0.48 
(0.549) 
0.32 
(0.463) 
1.10*** 
(0.318) 
-0.32 
(0.502) 
0.35 
(0.433) 
0.21 
(0.363) 
-0.14 
(0.618) 
-0.86* 
(0.499) 
-1.40*** 
(0.399) 
0.45 
(0.606) 
-1.12* 
(0.589) 
-1.77*** 
(0.495) 
-0.61 
(0.785) 
-0.95 
(0.736) 
-2.40*** 
(0.364) 
-0.25 
(0.612) 
0.21 
(0.517) 
-0.69*** 
(0.204) (0.236) (0.205) (0.256) (0.289) (0.370) (0.217) 
Observations 508 508 508 508 508 508 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 24: Topical Interest Logistic Regression Results 
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_1
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_2
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_3
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_4
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_5
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_6
No
Yes
Predicted Probabilities Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects topicqz4_7
Figure 24: Topical Interest Logistic Predicted Values 
28 
9.3.7 Willingness to Participate in Contest to Produce Short Radio Segments 
(1) 
Participate Contest 
b/se 
Program 0.01 
(0.037) 
MassComm 0.00 
(0.060) 
PR -0.01 
(0.047) 
Program × MassComm 0.04 
(0.063) 
Program × PR -0.02 
(0.063) 
Constant 1.95*** 
(0.031) 
Observations 508 
* p¡0.10, ** p¡0.05, *** p¡0.01 
Table 25: Participate in Contest Regression Results 
0 1 2
Predicted Values
Control
Program
Treatment
Journalism MassComm
PR
Treatment Effects Participate Contest
Figure 25: Participate in Contest Predicted Values 
29 
U.S. AGENCY FOR INTERNATIONAL DEVELOPMENT
1300 Pennsylvania Avenue, NW
Washington, DC 20523