1 
USAID/WEST AFRICA (USAID/WA) EVALUATION & ANALYTICAL 
SERVICES (EAS) PROJECT FOR THE REGIONAL PEACE AND 
GOVERNANCE PROGRAMS 
MIDLINE REPORT FOR IMPACT EVALUATION OF THE PEACE 
THROUGH DEVELOPMENT PHASE II (PDEV II) PROJECT IN CHAD, 
NIGER, AND BURKINA FASO 
Prepared By: 
Steven E. Finkel, Ph.D. 
Chris A. Belasco, Ph.D. 
Michael Neureiter 
John McCauley, Ph.D. 
Bruno Hoepers 
Caitlin C. Corrigan, Ph.D. 
November 2017 
This publication is made possible by the generous support of the American people through the United States Agency for International Development 
(USAID/West Africa) Evaluation & Analytical Services (EAS) for the West Africa Regional Peace and Governance Program (WARP) implemented by 
The Mitchell Group, Inc., Contract Number GS-10F-004N/AID-624-M-13-00001. The contents are the responsibility of the authors and do not necessarily 
reflect the views of USAID or the United States Government. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
i 
Table of Contents 
Table of Contents ............................................................................................................. i
Table of Figures ...............................................................................................................ii
Table of Tables ............................................................................................................... iii
Acknowledgments ........................................................................................................... v
Executive Summary ........................................................................................................ 1
I. Introduction ................................................................................................................ 12
A. Overview of Report ............................................................................................. 12
B. Description of the Peace through Development (P-DEV II) Program .................. 13
II. Survey Data Collection Overview .............................................................................. 14
B. Baseline and Midline Survey Data Collection ........................................................ 14
Data Collection Timeline ........................................................................................ 16
C. Sample Characteristics, Baseline and Midline Surveys ........................................ 16
D. Statistical Power ................................................................................................... 17
III. Results ..................................................................................................................... 19
A. P-DEV II Program Activities, 2013-2015 ............................................................. 19
B. Difference in Differences Analysis, Core versus Non-Core Zones ..................... 21
Goal 1: Social Cohesion ......................................................................................... 22
Goal 2: Resilience to Violent Extremism ................................................................ 30
Goal 3: Youth Outlook ............................................................................................ 43
C. The Impact of “Low Intensity” Versus “High Intensity” P-DEV II Programming ..... 51
Goal 1: Social Cohesion ......................................................................................... 52
Goal 2: Resilience to Violent Extremism ................................................................ 56
Goal 3: Youth Outlook ............................................................................................ 62
D. Panel Analysis ...................................................................................................... 65
Goal 1: Social Cohesion ......................................................................................... 69
Goal 3: Youth Outlook ............................................................................................ 70
E. Experimental Evidence .......................................................................................... 71
IV. Observations from Qualitative Data ......................................................................... 77
A. Burkina Faso ......................................................................................................... 77
Goal 1: Social Cohesion ......................................................................................... 79
Goal 2: Resilience to Violent Extremism ................................................................ 79
Goal 3: Youth Outlook ............................................................................................ 80
B. Niger ...................................................................................................................... 81
Goal 1: Social Cohesion ......................................................................................... 81
Goal 2: Resilience to Violent Extremism ................................................................ 82
Goal 3: Youth Outlook ............................................................................................ 82
C. Summary ............................................................................................................... 83
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
ii 
V. Conclusion ................................................................................................................ 83
VI. Appendix .................................................................................................................A-1
A. Methodology .....................................................................................................A-1
1. Survey Instrument .............................................................................................A-1 
2. Survey Data Collection Interviewer Training .....................................................A-2 
Procedures for “Fresh” Interviews .....................................................................A-2 
Panel Procedures and Verification ....................................................................A-2 
3. P-DEV II Activity Tracker Data Collection ..........................................................A-4 
4. Qualitative Data Collection ................................................................................A-5 
B. Questionnaire .......................................................................................................A-6 
B. Regression Tables for Panel Analysis ................................................................ A-57
C. Qualitative Data Collection Focus Group and Interviews ................................... A-67
Table of Figures 
Figure 1: Midline Statistical Power by Total Number of Communes, N=31 ................... 18
Figure 2: Interpersonal Trust ......................................................................................... 23
Figure 3: Institutional Trust ............................................................................................ 25
Figure 4: Community Decision-making .......................................................................... 26
Figure 5: Political Participation ...................................................................................... 28
Figure 6: Support for Interethnic Marriage ..................................................................... 29
Figure 7: Access to Jobs ............................................................................................... 32
Figure 8: Access to Vocational School .......................................................................... 33
Figure 9: Political Efficacy ............................................................................................. 34
Figure 10: Perceived Ethnic Differences ....................................................................... 36
Figure 11: Perceived Religious Differences .................................................................. 37
Figure 12: Religious Violence Is Justified ...................................................................... 38
Figure 13: Violence Is Effective to Solve Problems ....................................................... 39
Figure 14: Violence in the Name of Islam ...................................................................... 40
Figure 15: U.S. Is at War with Islam .............................................................................. 42
Figure 16: Life Satisfaction ............................................................................................ 44
Figure 17: Economic Outlook ........................................................................................ 45
Figure 18: Gender Inclusiveness of Middle School ....................................................... 46
Figure 19: Interest in Community Affairs ....................................................................... 48
Figure 20: Political Knowledge ...................................................................................... 49
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
iii 
Figure 21: Binary Activity Indicator for Niger ................................................................. 67
Figure 22: Binary Activity Indicator for Burkina Faso ..................................................... 67
Figure 23: Change in Program Participation ................................................................. 68
Table of Tables 
Table 1: Summary of Baseline Data Collection ............................................................. 15
Table 2: Summary of Midline Data Collection ............................................................... 16
Table 3: Demographic Profiles by Survey and Core Zone Status ................................. 17
Table 4: Average Number of Core Zone P-DEV II Activities Related to Each Strategic 
Objective ....................................................................................................................... 20
Table 5: Interpersonal Trust .......................................................................................... 23
Table 6: Institutional Trust ............................................................................................. 25
Table 7: Community Decision-making ........................................................................... 26
Table 8: Political Participation ....................................................................................... 28
Table 9: Interethnic Marriage ........................................................................................ 29
Table 10: Access to Jobs .............................................................................................. 32
Table 11: Access to Vocational School ......................................................................... 33
Table 12: Political Efficacy ............................................................................................ 34
Table 13: Perceived Ethnic Differences ........................................................................ 36
Table 14: Perceived Religious Differences .................................................................... 37
Table 15: Religious Violence Is Justified ....................................................................... 38
Table 16: Violence Is Effective to Solve Problems ........................................................ 39
Table 17: Violence in the Name of Islam ....................................................................... 40
Table 18: U.S. Is at War with Islam ............................................................................... 42
Table 19: Life Satisfaction ............................................................................................. 44
Table 20: Economic Outlook ......................................................................................... 45
Table 21: Gender Inclusiveness of Middle School ........................................................ 46
Table 22: Interest in Community Affairs ........................................................................ 48
Table 23: Political Knowledge ....................................................................................... 49
Table 24: Interpersonal Trust ........................................................................................ 53
Table 25: Institutional Trust ........................................................................................... 53
Table 26: Community Decision-making ......................................................................... 54
Table 27: Political Participation ..................................................................................... 55
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
iv 
Table 28: Support for Interethnic Marriage .................................................................... 55
Table 29: Access to Jobs .............................................................................................. 56
Table 30: Access to Vocational School ......................................................................... 57
Table 31: Political Efficacy ............................................................................................ 58
Table 32: Ethnic Differences ......................................................................................... 58
Table 33: Religious Differences .................................................................................... 59
Table 34: Religious Violence Is Justified ....................................................................... 59
Table 35: Violence Is Effective to Solve Problems ........................................................ 60
Table 36: Violence in the Name of Islam ....................................................................... 61
Table 37: Anti-West Attitudes ........................................................................................ 61
Table 38: Life Satisfaction ............................................................................................. 62
Table 39: Economic Outlook ......................................................................................... 63
Table 40: Inclusiveness of Schools ............................................................................... 63
Table 41: Interest in Community Affairs ........................................................................ 64
Table 42: Political Knowledge ....................................................................................... 64
Table 43: Panel Analysis for Goal 1: Social Cohesion .................................................. 69
Table 44: Panel Analysis for Goal 2: Resilience to Violent Extremism .......................... 70
Table 45: Panel Analysis for Goal 3: Youth Outlook ..................................................... 70
Table 46: List Experiment #1 – Percentage of Respondents Who Choose the Sensitive 
Item ............................................................................................................................... 72
Table 47: List Experiment #1 – Percentage of Respondents Who Choose the Sensitive 
Item ............................................................................................................................... 73
Table 48: List Experiment #2 – Percentage of Respondents Who Choose the Sensitive 
Item ............................................................................................................................... 75
Table 49: Endorsement Experiment #1 – Percentage of Midline Respondents Who 
Disapprove of Vaccination Program .............................................................................. 76
Table 50: Endorsement Experiment #1 – DiD Analysis ................................................. 76
Table 51: Summary of Results ...................................................................................... 85
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
v 
Acknowledgments 
The authors would like to acknowledge the contribution of the following members of the 
team in their efforts to produce this report. Without their support, this report would not 
have been possible. 
EAS Team Survey and Regional Specialist Social Scientist John McCauley, Ph.D. led 
survey training teams in Burkina Faso, and Niger, conducted quality control on piloting 
and early survey implementation in both countries, and conducted training of qualitative 
data collection teams at Centre d’Analyses et d’Actions Pour la Sécurité et la Paix 
(CASPA) in Niger and at Centre d’Etudes, de Recherches et de Formation pour le 
Développement Economique et Social (CERFODES) in Burkina Faso. EAS team fellow 
Douzounet Mallaye, Ph.D. led training in Chad. CERFODES technical lead Kadisso 
Nacambo led follow-on trainings in Niger and Chad. Senior Consultant for Research and 
Evaluation Robert Charlick, Ph.D. at The Mitchell Group contributed to the qualitative 
questionnaire and developed training materials for the qualitative specialists. 
Quality Control Specialists Adamou Hamadou, Aristide Mabali, and Ye Bintou Sanou 
conducted quality control for the respective survey teams in Niger, Chad, and Burkina 
Faso. 
We would also like to thank the USAID/ West Africa Mission Director and Regional Peace 
and Governance Office for comments on our preliminary results presentation and 
cooperation throughout the process of data collection; and the teams at CERFODES, 
CESEV, ATEP Tchad, and CASPA for their dedicated efforts in delivering quality data. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
A-1 
Executive Summary 
This report presents the results of the mid-line analysis of the impact of the Peace through 
Development (P-DEV II) program on a series of outcomes related to countering violent 
extremism and strengthening community resilience in Chad, Niger, and Burkina Faso. We 
use a combination of survey data, program activity trackers and quarterly reports, and 
qualitative data collected in all of the zones in which the P-DEV II program was active in 
the three countries. The evaluation consists of the analysis of data from baseline surveys 
conducted in each of the three countries in 2013 along with new interviews conducted in 
2015-2016 in order to determine whether P-DEV II activities led to changes in community 
and individual-level outcomes corresponding to the primary goals of the program: 
improving social cohesion, resilience to violent extremism, and improving the educational 
and employment outlook for youth. 
The basic design of the impact evaluation consists of comparisons between core zones 
and non-core zones in Chad, Niger, and Burkina Faso. Core zones are areas that were 
designated by International Relief and Development (IRD), the implementing entity for P￾DEV II, and USAID for exposure to the full array of P-DEV II program activities, while non￾core zones are areas that received or were expected to receive only the program’s media 
treatments, primarily radio programming related to good governance and countering 
violent extremism among youth. The program operated in a total of 83 zones in the three 
countries—40 core and 43 non-core. Baseline survey data was collected in 2013 on 7,720 
respondents in the 83 zones, following procedures described in Finkel et al. (2014).1
 Mid￾line survey data was collected in 2015-2016 on 2,577 respondents in the same 83 zones, 
with 2,080 representing “fresh” respondents and 497 being follow-up (or “panel”) 
participants who were also interviewed in the baseline wave of data collection. 
The longitudinal structure of the data allows for estimation of “difference-in-differences” 
(DiD) in program outcomes, that is, over time comparison of the changes in program￾relevant outcomes in core zones and changes in program-relevant outcomes in non-core 
zones from the baseline interviews to the midline. The DiD represents the estimated effect 
that can be attributed to the non-radio potion of P-DEV II programming, This basic 
comparison between core and non-core zones in terms of changes over time is then 
supplemented with more nuanced analyses. First, we make use of the quarterly P-DEV 
II activity trackers to determine which zones in each country received more and less 
extensive programmatic activities, and we attempt to ascertain whether the intensity of P￾DEV II activities is associated with stronger or weaker program impact. Second, we 
exploit the panel portion of the study to determine whether individual-level increases in 
exposure to activities led to changes in outcomes, i.e., whether individuals who reported 
increased exposure to different aspects of the program in the intervening periods of data 
collection showed greater impact than individuals who participated less in program 
activities. Third, we conduct analyses of program impact on measures of support for 
violent extremism that are gauged with innovative “list” and “endorsement” experimental 
methods designed to overcome potential social desirability biases associated with 
 
1
 Steven E. Finkel, Reynaldo Rojo Mendoza, Cassilde Schwartz, Chris Belasco, and Aaron Abbarno. 2014. “Baseline 
Report for Impact Evaluation of the Peace through Development Program II (P-DEV II) in Chad, Niger, and Burkina 
Faso.” Report submitted to USAID/West Africa. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
2 
responses to sensitive questions. Finally, we conducted an extensive series of qualitative 
interviews and focus groups in order to supplement the quantitative analyses and arrive 
at a more nuanced understanding of potential program impacts. 
Data Collection 
The survey instrument developed for the P-DEV II baseline study was used as the basis 
for the midline impact evaluation. Two survey instruments, one developed by IRD￾InterMedia and the other by the EAS team serve as the foundation for the evaluation. The 
original survey was developed by IRD-InterMedia in consultation with USAID personnel, 
and served as the instrument for the first wave of interviews conducted in Chad, Niger, 
and Burkina Faso between March and September 2013. The EAS evaluation team 
designed a second survey instrument that built on the one previously designed by IRD￾InterMedia in order to improve and extend the work that had already been done. The EAS 
survey instrument kept 65% of the items in the original IRD-InterMedia survey instrument 
in order to combine information from both samples into a single data set that enhances 
the quality of the evaluation. This survey was altered slightly for purposes of the mid-line 
data collection, with approximately 25 questions being eliminated in order to shorten 
interview time. 
We selected 19 indicators, or outcome measures, for analysis, each relating to one of the 
three specific and measurable goals of the P-DEV II program: 
Goal 1: Improvements in social cohesion through generation of the norms and 
networks that enable collective action as measured through groups and networks, 
increased trust within networks and among strangers, increased social inclusion, and 
improved ability of communities to communicate among each other and with other 
communities. 
Specific Indicators related to this goal include: Interpersonal Trust, Institutional Trust, 
Perceptions of Inclusive Community Decision-Making, Political Participation, and Support 
for Interethnic Marriage. 
Goal 2: Resilience to violent extremism through reduction of risk to vulnerable 
individuals who could become radicalized to the point of being willing to use violence by 
strengthening factors that enable vulnerable individuals to resist violent extremism. This 
includes attitudes toward violence and extremist ideologies, community leadership, social 
and political engagement. 
Specific Indicators related to this goal include: Perceived Access to Jobs, Perceived 
Access to Vocational Schools, Political Efficacy, Perceived Ethnic Differences, Perceived 
Religious Differences, Belief in the Justifiability of Religious Violence, Belief that Violence 
is Effective, Belief in the Justifiability of Violence in the Name of Islam, and Belief that the 
U.S. is at war with Islam. 
Goal 3: Improvements in youth outlook through the individual and collective vision of 
the future, in outlooks on: economic outcomes, participation in civil society and local 
decision making, attitudes toward existing and potential conflict in their societies, and 
expectations regarding the education and learning environment. 
Specific Indicators related to this goal include: Life Satisfaction, Economic Outlook, 
Perceived Gender Inclusiveness of Middle Schools, Interest in Community Affairs, and 
Political Knowledge. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
3 
The EAS team utilized its Burkina-based partner Centre d'Etudes, de Recherches et de 
Formation pour le Développement Economique et Social (CERFODES) for data 
collection, which worked with local partners Association Tchadienne pour l'Étude de la 
Population (ATEP) in Chad and Cabinet d’Expertise Suivi-Evaluation (CESEV) in Niger. 
The EAS Team conducted follow-up survey data collection in all 83 core and non-core 
zones in which baseline data was collected. The initial plan was to collect interviews with 
25 new (“fresh”) randomly selected respondents in all zones, along with follow-up data 
collection in the 35 EAS zones with 25 of the same respondents (“panel”) identified during 
the baseline wave per zone. Panel interviews were not possible in the IRD zones because 
of the lack of identifying information collected from respondents during the baseline phase 
(as follow-up interviews with the same individuals were not envisioned by IRD/USAID at 
that time). Unforeseen bureaucratic issues in Chad, however, led to a significant delay in 
the data collection, and it was decided to eliminate the “panel” portion of the study there 
in order to complete the midline analysis in a timely fashion. In all, 2,577 interviews were 
conducted in the midline phase. The table below shows the dates of data collection, the 
number of new interviews, panel interviews, and core and non-core zones per country. 
Summary of Midline Data Collection 
Data collection was led by CERFODES technical lead Ms. Nacambo and supervised by 
Dr. Belasco. Burkina Faso data collection began August 24, 2015, and was suspended 
between 17 September and 23 October following the coup d’état in the country. Data 
collection resumed 20 October with additional interviews collected through the week of 2 
November. In Niger data collection began August 31, 2016 concluded November 10, 
2016 after additional interviews were conducted to ensure balance among questionnaire 
types and gender. Unforeseen security issues prevented data collection from occurring 
at the same time in Chad. Data collection there occurred for fresh interviews only from 
March 21-April 3, 2016. 
Survey results showed that baseline and midline respondents were generally quite similar 
in terms of basic demographics, with the exception that midline respondents were slightly 
less religious (measured as the average number of days per week on which the 
respondent attends religious service) and somewhat more educated. These differences, 
however, were reflected in both core and non-core zones, so do not represent a significant 
confound in the statistical analyses. 
Qualitative data was collected by CERFODES in Burkina Faso and by project partner 
Cabinet d'Analyses et d'Actions pour la Sécurité et la Paix (CASPA) in Niger. Midline 
qualitative data collection included 10 focus groups and 12 interviews conducted in 
Country Date Zones # of Interviews
Core Non-core Core Non-core
Chad March – April 2016 15 14 376 349
Niger Aug. – Nov. 2015 15 16 360 (Fresh) 
129 (Panel)
420 (Fresh) 
121 (Panel)
Burkina 
Faso Aug. – Nov. 2015 10 13 252 (Fresh) 
101 (Panel)
323 (Fresh) 
146 (Panel)
40 43 1,218 1,359
Total Interviews 2,577
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
4 
Burkina Faso and 10 focus groups and 21 interviews conducted in Niger among a 
strategic selection of actors implementing the radio, community leadership, youth 
leadership strengthening, Imam and Quranic school training, and vocational training 
activities. Given the diverse nature of the recipients, these will be complemented by 
external review of the training and strengthening activities. 
The EAS team held separate focus groups for women, led by women, to encourage 
participation and free exchange of opinion. All materials including the interview and focus 
group field notes have been utilized to incorporate qualitative findings into the midline 
report. 
Analysis 
Scope of P-DEV II Activities 
P-DEV II activity trackers indicate that as of 3-31-2015, a total of more than 1,400 zone￾specific activities had been carried out, which corresponds to an average of about 33 
activities per zone. (National activities that affect all zones are excluded from these totals). 
The minimum value of activities implemented in a core zone is two, and the maximum 
value is 74. There are some noticeable differences between the three countries in terms 
of how many activities were carried out in each. In Chad, the mean value of activities 
implemented in the core zones is 23, whereas this figure is 27 for Niger and 60 for Burkina 
Faso. There are relatively more activities relating to P-DEV II strategic objectives #1 
(Youth Empowered) and #3 (Civil Society Capacity) compared to activities relating to 
strategic objectives #2 (Moderate Voices) and #4 (Strengthened Local Government).
These differences were generally consistent across the three country contexts. 
Estimation of P-DEV II Program Impact 
The basic analyses consisted of standard DiD regression models to estimate the impact 
of P-DEV II programming in core versus non-core zones. In equation form, the DiD model 
is expressed as: 
Yit = a + b1Corei + b 2Wavet + b 3Corei
*Wavet + e it
where Y (a given outcome) at a given point in time is equal to: a common intercept or 
starting point (α), an effect (β1) of whether the individual is in a core zone or not, an effect 
(β2) of a given wave of observation (baseline versus midline) on all individuals, an 
interaction effect (β3) of wave of observation with core zone status, and an idiosyncratic 
error term (εit). Taking differences in the above equation means that the non-core zones 
change over time is represented by β2, and the core zones change over time is β2+β3, 
with the “difference in differences” in the two groups being β3. This coefficient represents 
the DiD causal effect of the treatment. We estimate this model for the pooled (three￾country) sample in order to arrive at the overall DiD impact of P-DEV II, and we estimate 
an additional variant which allows country-by-country impacts to be estimated as well. 
Results 
We summarize the results of the quantitative analyses in the table below. We show 
positive significant effects of the program, i.e., in line with P-DEV II program goals, in 
green, and negative significant effects, i.e., in the “wrong” direction, in red. Effects that 
were registered in the pooled cross-country analysis are shown in bold faced capital 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
5 
letters. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
6 
Summary of Results 
Core vs. 
Non-core
Low Intensity 
Core vs. Non￾core
High Intensity 
Core vs. Non￾core
Panel 
Analysis 
Social Cohesion
Interpersonal Trust
Chad 
Niger 
Burkina
Institutional Trust Burkina
Community Decision￾Making Burkina Burkina Chad
Political Participation Burkina Niger
Interethnic Marriage Chad Chad 
POOLED Burkina
Resilience to Extremism
Access to Jobs
Chad 
Burkina
POOLED
Access to Vocational 
Schools
Chad 
Burkina Burkina Chad Burkina
Political Efficacy Burkina Burkina
Perceived Ethnic 
Differences Burkina Burkina Niger
Perceived Religious 
Differences Burkina Burkina Niger
Justifiability of 
Religious Violence Burkina POOLED Burkina
Violence is Effective Chad 
Burkina Niger
Justifiability of Violence 
in the Name of Islam Burkina Niger
U.S. is at war with 
Islam
Chad 
Niger 
Burkina 
POOLED
Chad 
Niger 
Burkina 
POOLED
Chad 
Niger 
Burkina
Niger
Youth Outlook
Life Satisfaction Niger
Economic Outlook Burkina
Chad 
Niger 
Burkina
Burkina
Inclusiveness of Middle 
Schools Chad Chad Niger
Interest in Community 
Affairs
Chad 
Niger 
Burkina
Political Knowledge
Niger 
Burkina 
POOLED
Niger 
Burkina
Niger 
Burkina 
POOLED
Niger 
Key: Positive Effect (Statistically Significant); Negative Effect (Statistically Significant) 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
7 
The results may be summarized as follows: 
When outcomes were pooled across all three countries, there were only two indicators of 
the 19 analyzed in which a statistically significant pooled DiD effect of core zone versus 
non-core zone status was found (see Section III-B). This means that on the 
overwhelming majority of outcomes relevant to the goals of the P-DEV II program, 
there was no detectable general difference across countries in the trends over time 
between core and non-core zones, although, as noted, the pooled effects may mask 
important country-specific effects. Positive pooled DiD effects were found on political 
knowledge (a two-item measure of respondent’s factual knowledge about the President’s 
length of term and the number of seats in the national legislature), and on the 
respondent’s perception that the US is not at war with Islam. 
An examination of the country-specific effects showed numerous significant impacts of 
the program, especially in Burkina Faso (see Section III-B). This may reflect the 
substantially greater number of P-DEV II activities carried out on average in Burkina core 
zones than in the two other countries. Significant DiD estimates of positive program 
impact were noted on eight indicators in Burkina Faso, five of which related to Program 
Goal 2, Resilience to Violent Extremism: access to vocational schools, political efficacy, 
reduced perceptions of ethnic and religious differences, and reduced justifiability of 
religious violence. Burkina core zones also increased more than non-core zones on 
perceptions of the inclusiveness of community decision-making, political 
knowledge, and general economic outlook, though core zones in Burkina 
increased more on the perception that the US is at war with Islam. 
The pattern of significant results suggests null findings in Chad and Niger. There were 
few significant DiD impacts in Chad or Niger (see Section III-B), with three significant 
positive impacts for core zones differences relative to non-core zones in Chad (support 
for interethnic marriage, decreased perception of the US at war with Islam, and increased 
support for gender inclusiveness in middle schools), and two in Niger (decreased 
perception of the US at war with Islam and political knowledge). 
Analysis of DiD impacts in “high intensity” zones – i.e., those with greater than average 
total P-DEV II activities – shows one additional significant pooled (cross-country) impacts 
of the program, on perceived access to jobs (see Section III-C). Differences over time on 
this indicator in the high-intensity core zones were greater than differences in the non￾core zones. Two additional indicators — the (un)justifiability of religious violence and 
support for interethnic marriage — show significant pooled effects in low-intensity zones 
compared to non-core zones, though in both cases these effects reduce to nearly zero in 
the high intensity P-DEV II zones. 
Country-by-country “high intensity” zone analysis showed a mixed pattern of impacts: 
greater numbers of indicators in Burkina were significant in the high-intensity zones, 
especially those pertaining to Resilience to Extremism (see Section III-C). Altogether 10 
of the 19 indicators showed significant DiD impacts in Burkina Faso when comparing high 
intensity core zones to non-core zones. At the same time, three indicators in those 
Burkina Faso zones also moved anomalously in the “wrong” direction, all related to 
Resilience to Violent Extremism. Similarly, in Chad and Niger there were sporadic positive 
impacts in high-intensity zones (4 in Chad, 2 in Niger) along with effects in the “wrong” 
direction (5 in Chad, 4 in Niger) in terms of the goals of the P-DEV II program. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
8 
Analysis of the panel portion of the data failed to replicate the core findings of the zone￾level analyses (see Section III-D), likely due to the small number of individuals in each 
zone who were re-interviewed from the baseline wave limiting our statistical power to 
detect small program effects, along with other difficulties in the panel portion of the data 
collection described earlier. The results did show that individuals’ experiences with 
activities related to countering violent extremism increased in the core zones 
compared to individuals in the non-core zones over time. But few of those changes 
were associated positively with changes in those same individuals’ attitudes and 
perception since the baseline interviews. Both positive and negative program effects 
are observed in the panel analysis in Niger and Burkina Faso, though overall pattern of 
results in panel analysis in the two countries probably reflect essentially null effects of the 
program in both countries. All of the impacts found were relatively small in substantive 
magnitude, with a few reaching significance in the positive direction and a few in the 
negative direction. Similar balance between “near significant” effects were found in the 
positive and negative direction as well. All of this suggests that, from the panel analysis, 
we obtain a picture of generally null overall impacts in Chad and Niger. 
The study included two “list experiments” and one “endorsement” experiment designed 
to measure in unobtrusive or indirect ways the respondent’s willingness to consider 
engaging in specific acts of violent extremism or express approval of groups which 
engage in violent acts (see Section III-E). The results of these analyses suggest that 
there is a sizable pool of individuals in all country contexts who would support “an 
attack on [a] Western country’s embassy that could result in military or civilian 
casualties” (list experiment #1), who would consider “joining a group that carries 
out acts of violence to defend your religion” (list experiment #2), or who would be 
more likely to oppose a polio vaccination program if they are told that “Al-Qaeda 
in the Islamic Maghreb (AQIM), an Islamist group, will oppose this program” 
(endorsement experiment #1). These figures range from approximately 10-25% of the 
sampled populations. In general, however, there were no detectable P-DEV II core zone 
impacts on these outcomes, in that respondents in the core zones and non-core zones 
showed similar trends in these outcomes over time.
Examination of key informant interviews and focus groups in Burkina Faso and Niger 
showed that attitudes relating to program goals have deteriorated in the countries since 
the 2013 baseline data collection, providing context relating to support for toward violent 
extremism. While P-DEV II has targeted communities in order to build resilience against 
violent extremism, the rise in violent incidents that has occurred since baseline data 
collection may lead to increased perceptions of legitimacy for violent extremism. This is 
particularly true in Niger where perceptions of program effects are weaker and Boko 
Haram has strengthened its presence. Community tensions exist in Burkina Faso in 
the struggle over land and natural resources and in inter-religious tolerance. In 
both countries, factors that leave youth feeling neglected also tend to facilitate 
extremist recruitment and draw young people into violence. Participants in the 
qualitative data collection noted positive influences from P-DEV II programs in their 
countries but suggest that programs of this sort must be expanded reach more people, 
and thus to combat the problems of frustration and limited opportunity that can undermine 
peace. 
The Context of Violence within the P-DEV II Program Area 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
9 
The P-DEV II program region encountered increased exposure to violence following the 
baseline data collection wave. According to the Armed Conflict and Local Event Data 
Project only Niamey, Niger was exposed to Islamist violence prior to the baseline. 2
However, by the midline data collection, nearly 20 percent of all communes were the 
targets of Islamist attacks. 
Between 2014 and 2016, target zones in Chad, Burkina Faso, and Niger were exposed 
to multiple attacks by Islamist groups such as Boko Haram and the Islamist State. Overall, 
1,330 individuals died as a result of violent Islamist incidents perpetrated in target zones. 
Nearly three quarters of those fatalities occurred in the Nigerien cities of Diffa, Bosso, and 
N’Guimi. While the Islamic State and other groups such as the Al Mourabitoune Battalion 
and Ansar Dine were active in the region, Boko Haram was responsible for nearly 99% 
of all fatalities and 86% of all incidents. Most attacks (78%) were targeted at military 
officials and other security forces, while in 22% of all incidents civilians were the direct 
targets of the attack. Diffa, Niger (29%), Bosso, Niger (28%) and Bol, Chad (18%) were 
the three target zones most vulnerable to Islamist attacks. 
In Burkina Faso, the cities of Djibo, Markoye, Ouagadougou, and Tongomayel were 
particularly vulnerable to Islamist attacks. The Al Mourabitoune Battalion backed by Al 
Qaeda in the Islamic Maghreb attacked military forces and tourists in both Ouagadougou 
and Djibo, while the Islamic State became active in the border town of Markoye, where 
they attacked military forces. Another Al-Qaeda-affiliated group, Ansaroul Islam, was 
active in Djibo and Tongomayel. Ansaroul Islam conducted attacks against individuals 
who had deserted the group, including a local imam. 
In Chad, more than three quarters of the attacks that took place between 2014 and 2016 
were targeted at civilians and security forces in the town of Bol, in the Mamdi Department 
of the Lac Region. Boko Haram was responsible for all attacks in the city. Boko Haram 
conducted indiscriminate attacks, including suicide bombings in a fish market and at the 
Bougama military post. Boko Haram occasionally used children to conduct their attacks 
against civilians. N’Djamena was also the target of multiple suicide bombings, targeted at 
police headquarters, the national police academy, but also the central market and most 
recently the U.S. embassy. 
In Niger, most attacks were concentrated in the cities of Diffa and Bosso. In both target 
zones, Boko Haram was the unique perpetrator of the attacks. The group frequently set 
fire to victims, homes, and vehicles, as well as markets. The group also resorted to suicide 
bombers to attack military convoys and other military posts. In several instances, Nigerien 
troops managed to repel Boko Haram’s attacks, and successfully killed several hundreds 
of the group’s members. 
Increased exposure to violence, particularly by Boko Haram in Chad and Niger, may 
affect whether changes in attitudes occurred as anticipated by the P-DEV II program. As 
the security situation deteriorated, worsening of attitudes seen in measures of institutional 
trust, life satisfaction, and political participation reflect the context of violence within the 
affected zones. Program activities were concentrated within certain zones countries 
where violence was most prevalent. While there is a possibility that violence and P-DEV 
 
2
 Armed Conflict Location and Event Data (ACLED) Project (Raleigh et al. 2010). 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
10 
II program implementation could be related to each other, both impact attitudes of 
resilience toward violent extremism. However, violence is not expected to overturn the 
program results. Endline analysis could control for violence in order to account for its 
effect when measuring program impact. 
Sahel Political Context at P-DEV II Midline 
Program midline evaluation coincided with election schedules in all three P-DEV II 
Countries, causing data collection to occur at a time of heightened political discourse. 
Data collection in Burkina Faso occurred August-November 2015. These activities were 
interrupted during the coup d’etat launched 16 September 2015, during which 11 people 
died and more than 250 people were injured in attacks and political protests. Data 
collection resumed after 23 September after CERFODES determined that it could safely 
resume operations. However, we found that people were equally willing after the coup 
occurred as before to respond to the items in our survey. With new elections originally to 
be held October 11, 2015 (rescheduled to November 29, 2015). 
Niger data collection occurred August-November 2015, three months prior to National 
Assembly and first-round presidential elections in the country. With unforeseen security 
issues delaying activity in Chad until March 2016, data collection concluded one week 
prior to the 10 April presidential election. 
Increased political discourse may affect attitudes seen in measures of political 
participation, political efficacy, and political knowledge. With increases found in political 
participation, political knowledge, and interest in community affairs are in each country, 
survey respondents may be responding to messaging by parties and leaders during and 
prior to the electoral cycles in the three countries. 
Implications for Future Work 
The overall pattern of results found in the report has several implications for future work. 
First, it is clear that the P-DEV II program did not have consistent impacts across the 
three country contexts, as there were detectable differences in the trends over time in 
pooled core versus non-core zones on only two indicators out of the 19 analyzed. Instead, 
there was significant country-by-country variation in program impact: program effects 
were strongest in Burkina Faso, where positive impacts were registered in some way 
(core versus non-core zone, high intensity core zones versus non-core) on a majority of 
the 19 indicators, with substantially fewer and weaker effects registered in Chad and 
Niger. At the same time there were several anomalous effects in the “wrong direction” in 
all countries. This means future work will probably need to conduct more detailed 
analyses on a country-by-country basis, which will require additional numbers of 
interviews within each county in order to approach requisite levels of statistical power. 
Second, while it is possible that significant effects would be uncovered in theoretically 
meaningful subgroups of the population (e.g., men versus women, young versus old, rich 
versus poor), the fact that relatively few impacts were seen in the overall individual country 
samples—especially in Chad and Niger—means that detecting these potential impacts 
will be extremely challenging, requiring again substantially more interviews in the endline 
data collection phase. Finally, the lack of meaningful or consistent results emerging from 
the panel portion of the analysis suggests that it may not be wise to continue these efforts 
in subsequent data collection waves, given the low payoff relative to the high cost of 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
11 
collecting data on the same individuals over time. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
12 
I. Introduction 
A. Overview of Report 
This report presents the results of the mid-line analysis of the impact of the Peace through 
Development (P-DEV II) program on a series of outcomes related to countering violent 
extremism and strengthening community resilience in Chad, Niger, and Burkina Faso. We 
use a combination of survey data, program activity trackers and quarterly reports, and 
qualitative data collected in all of the zones in which the P-DEV II program was active in 
the three countries. The evaluation consists of the analysis of data from baseline surveys 
conducted in each of the three countries in 2013 along with new interviews conducted in 
2015-2016 in order to determine whether P-DEV II activities led to changes in community 
and individual-level outcomes corresponding to the primary goals of the program: 
improving social cohesion, resilience to violent extremism, and improving the educational 
and employment outlook for youth. 
The basic design of the impact evaluation consists of comparisons between core zones 
and non-core zones in Chad, Niger, and Burkina Faso. Core zones are areas that were 
designated by IRD (the implementing entity for P-DEV II) and USAID for exposure to the 
full array of P-DEV II program activities, while non-core zones are areas that received or 
were expected to receive only the program’s media treatments, primarily radio 
programming related to good governance and countering violent extremism among youth. 
The program operated in a total of 83 zones in the three countries—40 core and 43 non￾core. Baseline survey data was collected in 2013 on 7,720 respondents in the 83 zones, 
following procedures described in Finkel et al. (2014). 3
 Mid-line survey data was 
collected in 2015-2016 on 2,577 respondents in the same 83 zones, with 2,080 
representing “fresh” respondents and 497 being follow-up (or “panel”) participants who 
were also interviewed in the baseline wave of data collection. 
The longitudinal structure of the data allows for estimation of “difference-in-differences” 
(DiD) in program outcomes, that is, over time comparison of the changes in program￾relevant outcomes in core zones and changes in program-relevant outcomes in non-core 
zones from the baseline interviews to the midline. The DiD represents the estimated effect 
that can be attributed to the non-radio portion of P-DEV II programming, and these results, 
presented in Section III-B below, are the core findings of this report. This basic 
comparison between core and non-core zones in terms of changes over time will then be 
supplemented with more nuanced analyses. First, we make use of the quarterly P-DEV 
II activity trackers to determine which zones in each country received more and less 
extensive programmatic activities, and we attempt to ascertain whether the intensity of P￾DEV II activities is associated with stronger or weaker program impact (Section III-C). 
Second, we exploit the panel portion of the study to determine whether individual-level 
increases in exposure to activities led to changes in outcomes, i.e., whether individuals 
who reported increased exposure to different aspects of the program in the intervening 
periods of data collection showed greater impact than individuals who participated less in 
program activities (Section III-D). Third, we conduct analyses of program impact on 
 
3
 Steven E. Finkel, Reynaldo Rojo Mendoza, Cassilde Schwartz, Chris Belasco, and Aaron Abbarno. 2014. “Baseline 
Report for Impact Evaluation of the Peace through Development Program II (P-DEV II) in Chad, Niger, and Burkina 
Faso.” Report submitted to USAID/West Africa. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
13 
measures of support for violent extremism that are gauged with innovative “list” and 
“endorsement” experimental methods designed to overcome potential social desirability 
biases associated with responses to sensitive questions (Section III-D). Finally, we 
conducted an extensive series of qualitative interviews and focus groups in order to 
supplement the quantitative analyses and arrive at a more nuanced understanding of 
potential program impacts. A description of the qualitative data collection and findings can 
be found below in Section IV. 
B. Description of the Peace through Development (P-DEV II) Program 
The Peace through Development II (P-DEV II) program is a multi-year development 
program whose main objective is to counter extremist violence and adherence to 
extremist ideologies in Chad, Niger, and Burkina Faso. P-DEV II is the largest United 
States Agency for International Development/West Africa (USAID/WA)-funded program 
specifically designed for this purpose. P-DEV II activities are organized and structured to 
maximize the impact of the program in achieving its purpose. Under the overarching 
framework of countering violent extremism (CVE) through social and political 
development, P-DEV II has four multi-faceted, strategic objectives (SO): 
SO 1: Youth More Empowered through expanded livelihoods, vocational and 
entrepreneurial skills training, civic education, capacity building for youth associations, 
and leadership training to increase participation in local decision making by young men 
and women; 
SO 2: Moderate Voices Increased through integrated radio, social media, civic 
education, and conflict resolution activities, enhanced quality and credible information, 
and positive dialogue; 
SO 3: Civil Society Capacity Increased through formal and informal training, 
strengthened advocacy skills, citizen-led accountability initiatives and issue-based 
campaigns integrated with radio and social media and enhanced through civil society 
organizations (CSO) coalitions and networks; 
SO 4: Local Government Strengthened through organized and enhanced community 
entities and CSO capacity, greater citizen participation, and training in public 
administration, transparency, advocacy, and government outreach, and integrated with 
radio and social media. 
P-DEV II seeks to advance these four SO by concentrating its efforts on three more 
specific and measurable goals: 
Goal 1: Improvements in social cohesion through generation of the norms and 
networks that enable collective action as measured through groups and networks, 
increased trust within networks and among strangers, increased social inclusion, and 
improved ability of communities to communicate among each other and with other 
communities. 
Goal 2: Resilience to violent extremism through reduction of risk to vulnerable 
individuals who could become radicalized to the point of being willing to use violence by 
strengthening factors that enable vulnerable individuals to resist violent extremism. This 
includes attitudes toward violence and extremist ideologies, community leadership, social 
and political engagement. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
14 
Goal 3: Improvements in youth outlook through the individual and collective vision of 
the futures, in outlooks on: economic outcomes, participation in civil society and local 
decision making, attitudes toward existing and potential conflict in their societies, and 
expectations regarding the education and learning environment. 
The expectation is that furthering these goals would provide the foundation for 
communities that are more united by tolerance rather than extreme ideology, that are less 
likely to experience extremist violence, and that provide a more promising future for its 
members. Taken together, the attainment of these goals would build stronger and more 
resilient communities, which would constitute a powerful deterrent against violent 
extremism, extremist ideologies, and support for terrorist groups. 
P-DEV II conducts a range of activities in order to accomplish the program goals of social 
cohesion, resilience to violent extremism, and improved youth outlook. Project partners 
Search for Common Ground, The Salam Institute, and Equal Access subcontract 
implementation for these activities from IRD. 
P-DEV II activities that support the strategic objective of youth empowerment include: 
vocational training and the provision of livelihood assistance; literacy training for adults; 
provision of grants and in-kind resources to community schools; training youth in 
leadership and the conduct of conflict resolution activities such as participatory theater 
and media production; and support for community events with youth participation. 
P-DEV II activities that support the strategic objective of increased moderate voices 
include: Media outlets trained and assistance provided to radio stations in order to 
establish or improve broadcast capabilities; the distribution of public information through 
campaigns; the production of radio shows with themes of peace and tolerance; and the 
training of imams and the facilitation of intra-faith interreligious dialogue activities. 
P-DEV II activities that support the strategic objective of increased civil society capacity 
to address community issues include: civil society training; citizen advocacy training, and 
the formation of Community Action Committees; and support for community events. 
P-DEV II activities that support the strengthening local government include: Local 
government official capacity building; provision of grants and in-kind assistance to 
communities in support of development outcomes. 
We will discuss the nature and scope of the P-DEV II activities conducted between 
baseline and midline data collection in more detail in the results (Section III-A) below. 
II. Survey Data Collection Overview 
This section contains summary details for the data collection periods for the baseline and 
midline survey data collection, sample characteristics and statistical power for the Midline 
analysis. Details regarding the survey data collection instrument, enumerator training and 
panel verification procedures as well as descriptions of data collection procedures for 
qualitative and activity tracker analysis are included in Appendix A. 
B. Baseline and Midline Survey Data Collection 
Baseline data collection was conducted in core and non-core zones in Chad, Niger, and 
Burkina Faso. In 2013, two surveys were conducted in each country, one by IRD in 
conjunction with InterMedia, and one by the EAS Team. IRD-InterMedia went to the field 
for data collection in March 2013 for Chad and Niger, and in September 2013 for Burkina 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
15 
Faso. The first interviews for the EAS Team were conducted in Chad on September 16 
and the field period lasted until November 16. In Burkina Faso the data collection began 
on September 30 and the field period lasted until October 12. Due to a delay in Niger, the 
first official interviews occurred on November 10 and the field period lasted until 
November 30. 
IRD-InterMedia baseline data collection was conducted in 15 target zones (8 core, 7 non￾core) in Chad, 20 target zones (10 core, 10 non-core) in Niger, and 13 target zones (7 
core, 6 non-core) in Burkina Faso. The EAS Team baseline data collection was conducted 
in 15 target zones (7 core, 8 non-core) in Chad, 10 target zones (5 core, 5 non-core) in 
Niger, and 10 target zones (3 core, 7 non-core) in Burkina Faso. This brings total baseline 
data collection to 83 target zones (40 core, 43 non-core): 30 in Chad, 30 in Niger, and 23 
in Burkina Faso. Table 1 summarizes the baseline data collection timeline, the number of 
sampled zones, and number of interviews conducted per country and zone. 
Table 1: Summary of Baseline Data Collection 
The EAS team utilized its Burkina-based partner Centre d'Etudes, de Recherches et de 
Formation pour le Développement Economique et Social (CERFODES) for data 
collection, which worked with local partners Association Tchadienne pour l'Étude de la 
Population (ATEP) in Chad and Cabinet d’Expertise Suivi-Evaluation (CESEV) in Niger. 
The EAS Team conducted follow-up survey data collection in all 83 core and non-core 
zones in which baseline data was collected. The initial plan was to collect interviews with 
25 new (“fresh”) randomly selected respondents in all zones, along with follow-up data 
collection in the 35 EAS zones with 25 of the same respondents (“panel”) identified during 
the baseline wave per zone. Panel interviews were not possible in the IRD zones because 
of the lack of identifying information collected from respondents during the baseline phase 
(as follow-up interviews with the same individuals were not envisioned by IRD/USAID at 
that time). Unanticipated security issues in Chad, however, led to a significant delay in 
the data collection, and it was decided to eliminate the “panel” portion of the study there 
in order to complete the midline analysis in a timely fashion. In all, 2,577 interviews were 
conducted in the midline phase. Table 2 shows the dates of data collection, the number 
of new interviews, panel interviews, and core and non-core zones per country. 
Survey/Country Date Zones # of Interviews
Core Non-core Core Non-core
IRD￾InterMedia
Chad March 2013 8 7 640 560
Niger March 2013 10 10 819 790
Burkina Faso September 2013 6 7 481 560
24 24 1,940 1,910
EAS 
Chad Sept.-Oct. 2013 7 8 774 881
Niger November 2013 5 5 551 550
Burkina Faso Sept.-Oct. 2013 4 6 445 669
16 19 1,770 2,100
Total Interviews 3,710 4,010
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
16 
Table 2: Summary of Midline Data Collection 
While baseline data collection totaled 7,720 interviews with an average sample size of 93 
interviews per target zone, the midline contained only 25 interviews in IRD zones and 50 
interviews in EAS zones (25 new, 25 panel) due to budget constraints and to optimize 
statistical power. Due to budget considerations, midline data collection maximized 
available resources to the extent possible without sacrificing the rigor of the evaluation. 
Data Collection Timeline 
Data collection was led by CERFODES technical lead Ms. Nacambo and supervised by 
Dr. Belasco. Burkina Faso data collection began August 24, 2015, and was suspended 
between September 17 and October 23 following the coup d’état in the country. Data 
collection resumed October 20 with additional interviews collected through the week of 
November 2. In Niger data collection began August 31, 2015 and concluded November 
10, 2016 after additional interviews were conducted to ensure balance among 
questionnaire types and gender. Administrative issues associated with authorization for 
data collection from the Chadian Department of NGOs prevented data collection from 
occurring at the same time in Chad. 
C. Sample Characteristics, Baseline and Midline Surveys 
In this section, we present a basic demographic profile of the three countries under study. 
Table 3 illustrates descriptive statistics for our sample, broken down by baseline/midline 
survey and non-core/core zone. The table shows that the sample is evenly split between 
men and women as well as youth and non-youth (the latter being defined as respondents 
above the age of 30). For these two variables, gender and age, the differences between 
baseline and midline survey as well as between core and non-core zones are therefore 
negligible. In addition, baseline and midline respondents exhibit little difference in terms 
of poverty. To measure poverty, we created an additive index denoting how many out of 
twelve possible household items such as a fridge, TV, and radio respondents have in their 
household. Baseline and midline respondents exhibit little difference in terms of poverty, 
with respondents owning between two and three of these items. For both the baseline 
and midline survey, respondents in the core zones are slightly richer than those in the 
non-core zones. In addition, respondents in the core zones are somewhat more educated 
(measured as the percentage of respondents who are illiterate, and the percentage of 
respondents who have at least completed primary education) than those in the non-core 
Country Date Zones # of Interviews
Core Non-core Core Non-core
Chad March – April 2016 15 14 376 349
Niger Aug. – Nov. 2015 15 16 360 (Fresh) 
129 (Panel)
420 (Fresh) 
121 (Panel)
Burkina 
Faso Aug. – Nov. 2015 10 13 252 (Fresh) 
101 (Panel)
323 (Fresh) 
146 (Panel)
40 43 1,218 1,359
Total Interviews 2,577
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
17 
zones, a pattern that can be observed for both the baseline and the midline. Overall, then, 
the baseline and midline respondents are quite similar in terms of basic demographics, 
with the exception that midline respondents are slightly less religious (measured as the 
average number of days per week on which the respondent attends religious service) and 
somewhat more educated. These differences, however, are reflected in both core and 
non-core zones. 
Table 3: Demographic Profiles by Survey and Core Zone Status 
Baseline Survey Midline Survey
Non-core Core Non-core Core
Male (%) 52.4 50.1 53.2 52.0
Age (mean) 34.1 32.9 35.2 34.8
Youth (%) 48.5 52.2 43.3 46.8
Illiteracy (%) 36.7 34.3 38.6 28.5
Primary + (%) 23.5 34.5 29.9 43.9
Religious attendance per 
week 5.1 5.2 4.1 4.2 
Household Items (mean) 2.1 2.8 2.4 3.2
D. Statistical Power 
For purposes of the evaluation, statistical power refers to the probability that, if the true 
effect of P-DEV II in core zones is of a given size, we will find statistically significant effects 
on the outcomes of interest. Statistical power is closely related to the sample size and to 
the number of “treatment” and “control” units included in the study (i.e., core/non-core 
zones or communes). As the sample size and the number of zones increases, and the 
lower the variance of the estimated effect of the program, the higher the power. 
IRD-InterMedia and the EAS team collected baseline data in a total of 83 communes in 
Chad, Niger, and Burkina Faso (see Table 1 above). A total of 7,720 interviews were 
conducted with an average sample size of 93 interviews per commune. The study 
achieved high statistical power (>.90), given the large number of communes included in 
the study, even assuming relatively variable effect sizes across communes. 
For the midline evaluation, the same number of communes from the baseline study (83 
total) and an average sample size of 31 respondents per zone yielded lower but still 
acceptable levels of power. Figures 1 displays this graphically, with power generally 
greater than .90 depending on assumptions about effect size and variability across zones. 
It should be noted, however, that the statistical power of the study is lower when 
considering country-specific effects, given that the number of zones per county ranges 
from 23 to 30. It is also the case that the relatively small number of interviews per zone is 
not sufficient to allow detailed and reliable sub-group analysis, for example comparing 
men and women, youth and non-youth, etc. The number of interviews per zone in the 
midline phase was curtailed due to budgetary constraints, with the likelihood that a larger 
number of respondents will be interviewed in the endline phase, where more extensive 
subgroup analysis will be conducted. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
18 
Figure 1: Midline Statistical Power by Total Number of Communes, N=31 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
19 
III. Results 
A. P-DEV II Program Activities, 2013-2015 
We begin by providing information on the number and types of activities implemented in 
Chad, Niger, and Burkina Faso as part of the P-DEV II program. These figures were 
obtained from the Activity Trackers submitted by IRD and described in more detail above 
(Section II-E). As of 3-31-2015, a total of more than 1,400 activities had been carried out 
in the 43 core zones of the program (though of those, only 40 were surveyed for the 
midline), which corresponds to an average of about 33 activities per zone. The minimum 
value of activities implemented in a core zone is two, and the maximum value is 74. There 
are some noticeable differences between the three countries in terms of how many 
activities were carried out in each. In Chad, the mean value of activities implemented in 
the core zones is 23, whereas this figure is 27 for Niger and 60 for Burkina Faso. 
Next, we break down the scope of the P-DEV II by type of activity. The activities carried 
out through the P-DEV II program are classified into twelve different categories. Each of 
these types relates to one of the four strategic objectives (SO): 
SO #1: Youth more empowered 
1.1. Expanded youth livelihoods 
1.2. Increased access to education 
1.3. Strengthened youth leadership 
1.4. Increased youth mobilization 
SO #2: Increased moderate voices 
1.5. Increased capacity of media outlets 
1.6. Increased access to quality information 
1.7. Increased positive dialogue with religious leaders 
SO #3: Increased civil society capacity 
1.8. Increased CSO capacity 
1.9. Increased citizen participation 
SO #4: Strengthened local government 
1.10. Improved local government capacity 
1.11. Increased transparency and accountability in local decision-making 
1.12. More participatory local development 
Table 4 shows the average number of activities within each category implemented in the 
core zones. It can be seen that there are relatively numerous activities relating to strategic 
objectives #1 (Youth Empowered) and #3 (Civil Society Capacity), whereas those relating 
to strategic objectives #2 (Moderate Voices) and #4 (Strengthened Local Government) 
are relatively fewer. For example, in the average core zone, between three and four 
activities falling under category 1.3 “strengthened youth leadership” (such as participatory 
theater and mobile cinema) were carried out. Similarly, the average core zone saw the 
implementation of about six activities falling under category 1.4 “increased youth 
mobilization (such as the provision of sports equipment for municipalities, and youth-led 
awareness campaigns to promote peace/social cohesion). In stark contrast, only an 
average between one and two activities falling under category 2.3 “increased positive 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
20 
dialogue with religious leaders” (such as imam training on peace and non-violence) were 
carried out in the core zones. 
In addition, Table 4 reveals some interesting differences between the three countries. In 
Burkina Faso, each category saw the implementation of a number of activities greater 
than the pooled average, with the exception of category 2.3 “increased positive dialogue 
with religious leaders.” Chad exhibits a concentration of activities relating to strategic 
objective #1 (Youth Empowered), and is the only country where a great number of 
category 2.3 “increased positive dialogue with religious leaders was implemented. In 
Niger, a activities were consolidated under large sub-awards at the regional level, 
resulting in a count of fewer commune-level activities in strategic objective #4 
(Strengthened Local Government) were carried out. 
Table 4: Average Number of Core Zone P-DEV II Activities Related to Each Strategic Objective 
Pooled Chad Niger Burkina Faso
SO #1: Youth Empowered
1.1 Livelihood 3.116 1.857 1.158 8.600
1.2 Education 4.605 5.286 2.474 7.700
1.3 Leadership 3.814 2.500 4.000 5.300
1.4 Mobilization 6.000 4.857 5.947 7.700
SO #2: Increased Moderate Voices
 2.1 Media 
capacity 1.256 .286 2.000 1.200 
2.2 Quality info 1.302 .714 1.632 1.500
2.3 Imam training 1.674 2.643 1.368 .900
SO #3: Civil Society Capacity
3.1 CSO capacity 2.744 .429 1.632 8.100
3.2 Participation 4.605 1.714 5.105 7.700
SO #4: Local Government
4.1 Local capacity 1.442 .071 .842 4.500
4.2 Transparency 1.558 1.000 .421 4.500
4.3 Development 1.163 1.429 .579 1.900
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
21 
B. Difference in Differences Analysis, Core versus Non-Core Zones 
This section presents an analysis of the impact of the P-DEV II program on a series of 
outcomes related to the program goals of social cohesion, resilience against violent 
extremism, and youth outlook. The goals are operationalized according to numerous 
indicators that, taken together, identify the main components of the program’s strategic 
objectives. “Impact” refers to the differences in responses in the core zones between 
baseline and midline data collection compared to the differences in responses in non￾core zones during the same time period (hence the term “difference in differences” to 
describe this kind of analysis). For the moment, we ignore differences in the amount of 
activities conducted across zones (what may be called “treatment intensity”), and focus 
only on the differences between core zones, those targeted for the full range of P-DEV 
activities, and non-core zones, where at most only P-DEV-II radio programming was 
implemented. 
Our analysis proceeds in two steps. First, we present in a figure the distribution of each 
indicator in wave 1 and wave 2 zones within each country. Next, we present the results 
of statistical models to determine whether or not there are significant effects of the 
program on each indicator. These effects are calculated via the DiD approach: we regress 
goal-level indicators on dichotomous variables for core/non-core zones, wave 1/2, and 
country, as well as the interactions between these variables. If the program had the 
intended effects, we would see an improvement in core zones between wave 1 and 2 
relative to the development of the indicator in non-core zones between wave 1 and 2. 
In equation form, the DiD model is expressed as: 
(1) Yit = a + b1Corei + b 2Wavet + b 3Corei
*Wavet + e
it
where Y (a given outcome) at a given point in time is equal to: a common intercept or 
starting point (α), an effect (β1) of whether the individual is in a core zone or not, an effect 
(β2) of a given wave of observation (baseline versus midline) on all individuals, an 
interaction effect (β3) of wave of observation with core zone status, and an idiosyncratic 
error term (εit). For individuals at each point in time, this equation reduces to: 
(2) Non-Core, Baseline: Yi0 = a + e i0
 Non-Core, Midline: Yi1 = a + b 2 + e i1
 Core, Baseline: Yi0 = a + b1 + e
i0
 Core, Midline: Yi1 = a + b1 + b 2 + b 3 + e i1
Taking differences means that the non-core zones change over time is represented by β
2, and the core zones change over time is β2+β3, with the “difference in difference” in 
the two groups being β3. This coefficient represents the DiD causal effect of the 
treatment. 
We present the findings for each goal-level indicator exactly in line with these equations. 
The first row illustrates the average value for the respective indicator in each country’s 
non-core zones during the first wave of the survey, i.e. before the implementation of the 
P-DEV II program (α). The second row shows the average difference between non-core 
and core zones for wave 1, and whether this difference is statistically significant (β1). In 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
22 
the third row, we present the average difference between non-core zones for wave 1 and 
non-core zones for wave 2, and whether this difference is statistically significant (β2). 
This row can be understood as general developments in each country with respect to the 
indicator in question that are independent of the implementation and effects of the P-DEV 
II program. 
The fourth row is the one in which the actual effects of the P-DEV II program are shown. 
Here, we take the difference between wave 2 core zones and wave 1 core zones and 
subtract the difference between wave 2 non-core zones and wave 1 non-core zones (i.e., 
β3). If the value in the fourth row is statistically significant, it suggests that the change of 
the indicator in the core zones between waves is significantly different from the change 
in the non-core zones, which can be attributed to the implementation of the P-DEV II 
program. In addition to the effects for each country, the fourth row also contains a pooled 
treatment effect, which shows whether the program had a DiD effect when pooling all 
observations all three countries. For the reader’s convenience, the fourth row is color￾coded: coefficients representing a desired effect of the program are shown in green (i.e., 
effects leading to greater resistance to extremism, more positive outlooks, more social 
cohesion), whereas coefficients representing a detrimental effect of the program (less 
resistance, less positive outlooks, less social cohesion) are shown in red. 
The models were run first for the pooled sample, which means that all countries were 
combined into a single analysis and the difference in differences for all core versus non￾core P-DEV II zones was estimated, while allowing for country-specific initial levels and 
overall trends on the outcome between the baseline and endline. These results are shown 
as the “pooled” treatment effect in the tables that follow. We then interacted all of the 
variables in equation (1) by indicator variables corresponding to the specific countries so 
that, in essence, we estimate one model that characterizes the causal processes in the 
three different country contexts. We present the results of these analyses in separate 
columns for each country in the tables that follow. Standard errors for all regression 
coefficients are clustered by target zone, following standard procedures for grouped data 
of this kind. 
Goal 1: Social Cohesion
Figures 2 through 6 as well as Tables 5 through 9 present the results for the series of 
questions related to Goal 1: Social Cohesion. Social cohesion is a broad concept that is 
captured through two separate indicators: 1) interpersonal and institutional trust; and 2) 
social inclusiveness in the community. 
In summary, our DiD analysis reveals that few consistent positive social cohesion effects 
could be attributed to P-DEV II. There were no statistically significant pooled DiD 
treatment effects across the three countries for any of these indicators. Of five indicators 
of social cohesion, only interethnic marriage showed DiD effects that could be attributed 
to P-DEV II programming in the core versus non-core zones in Chad, and no DiD effects 
could be attributed to the program in Niger. Perceptions of ordinary people’s involvement 
in community decision-making declined significantly less in P-DEV II core zones in 
Burkina Faso than in non-core zones from baseline to midline. 
Interpersonal and Institutional Trust (higher values represent increased trust) 
To measure interpersonal trust, respondents were asked whether they disagree or agree 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
23 
that “most people are willing to help if you ask for help.” Figure 2 breaks down responses 
by country and wave. The figure shows that levels of interpersonal trust remained rather 
stable in Chad in Niger, while there appears to have been a noticeable drop in 
interpersonal trust in Burkina Faso, as more respondents overall disagreed with the 
statement. 
Focus groups from Burkina Faso suggested that this drop in interpersonal trust could be 
a function of trends in migration and movement in the region. In Seytenga, participants 
noted that people (and especially younger people) often have difficulty accepting those 
perceived as outsiders, and patterns in climate change may be increasing the share of 
people perceived as such as they move to find better agricultural opportunities. Further, 
there is some concern—expressed, for example, in a focus group in Yatenga—that theft 
is becoming increasingly common, which in turn may be undermining interpersonal trust. 
These patterns may be viewed as broader social and environmental issues in the region 
that are creating greater challenges, even as the PDEV II initiative works to combat the 
consequences. 
Figure 2: Interpersonal Trust 
Table 5: Interpersonal Trust 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.295 2.298 2.616 X
Baseline Core Difference .049 .025 -.229*** X
Baseline-Midline 
Difference, Non-Core
-.016 .136 -.308*** X 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Disagree Neither agree nor disagree
Agree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
24 
Difference in Differences, 
Core versus Non-Core -.058 -.136 .163 -.032 
* p < .10; ** p < .05; *** p < .01 
Table 5 shows that overall, there are no impacts in the table attributable to the P-DEV II 
program in terms of increasing interpersonal trust. Comparing country estimates of the 
treatment effect in core and non-core zones in Chad, Niger, Burkina, and the overall 
pooled treatment effect, we find no statistically significant differences over time. 
Interpersonal trust in Burkina has declined between two waves; this decline is shown by 
the -.308 significant coefficient in row 3 of the table. While there was an overall decline in 
institutional trust in the country, in core zones there was a positive (but statistically 
insignificant) change in core zones. Using difference in differences analysis, we find no 
statistically significant differences in the core Burkina zones, indicating that the general 
decline in trust occurred in statistically indistinguishable ways in both core and non-core 
communes. 
Institutional trust is measured through five individual trust items. Each respondent was 
presented with five separate statements: a) I trust local authorities; b) I trust the central 
government; c) I trust religious leaders; d) I trust NGOs; e) I trust the police. For each of 
these statements, they could choose among three options: disagree, neither, or agree. 
Figure 3 graphs the results of an additive index of the institutional trust items. It groups 
respondents into three categories depending on the number of institutions for which they 
chose the option “agree.” In all three countries, there appears to have been little change 
over time. 
Based on feedback from interviews and focus groups, it appears that attitudes regarding 
institutional trust are fairly entrenched and will take some time to adjust. As the President 
of a Women’s Association in Niamey explained in an interview: “government actors are 
perceived as regularly exploiting people, and then people are faced with either confronting 
government and coming into conflict with leaders or going along with government and 
thus losing their own trustworthiness.” This vicious cycle may explain why institutional 
trust has not changed over a two year period, despite the introduction of PDEV II activities. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
25 
Figure 3: Institutional Trust 
Table 6: Institutional Trust 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.487 2.589 2.737 X
Baseline Core Difference -.152* -.119* -.251*** X
Baseline-Midline 
Difference, Non-Core -.008 .009 -.053 X 
Difference in Differences, 
Core versus Non-Core -.026 .040 .070 .028 
* p < .10; ** p < .05; *** p < .01 
The analyses in Table 6 are based on respondents’ average level of trust across the five 
institutions. We find that the P-DEV II program had no discernible impact on respondents’ 
level of institutional trust: the table shows that there has been little change in institutional 
trust between baseline and midline. The difference in differences analysis comparing core 
and non-core zones shows slight positive changes in Niger Burkina, and in the pooled 
treatment effect, but these positive changes occurred in statistically indistinguishable 
ways in both core and non-core communes. 
Social Inclusiveness (higher values represent increased inclusiveness) 
One component of social inclusiveness is the involvement of members of the community 
in decision-making processes. To probe this dimension, respondents were asked to what 
extent ordinary people from the commune/neighborhood participate when important 
decisions are being made. As Figure 4 shows, the percentage of respondents who 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Trusts 0 or 1 institution Trusts 2 or 3 institutions
Trusts 4 or 5 institutions
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
26 
choose the option “a lot” has decreased in all three countries, suggesting a declining 
sense of involvement in such processes. 
Figure 4: Community Decision-making 
Table 7: Community Decision-making 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.947 2.190 2.518 X
Baseline Core 
Difference .037 -.035 -.405*** X 
Baseline-Midline 
Difference, Non-Core -.102 -.170 -.404*** X 
Difference in 
Differences, Core versus 
Non-Core
-.189 -.088 .323*** -.006 
* p < .10; ** p < .05; *** p < .01 
Table 7 provides a more nuanced picture of this development. In Chad, Niger, and in the 
overall pooled treatment effect estimate the differences between core and non-core, and 
between baseline and midline reflect this overall decline but are not statistically 
significant. 
In Burkina Faso, however, responses in core zones show that P-DEV II programming 
offsets the general decline in perceptions that ordinary people participate in the decision￾making processes in the community. While we find a significant decrease in perceived 
influence in community decision making in the non-core zones from baseline to midline, 
respondents in core zones showed significantly less negative changes over time. In other 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Not at all Somewhat
A lot
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
27 
words, non-core zones decreased between baseline and midline on average by -.404, 
while core zones decreased by .081 (-.404 + .323= -.081), with this difference in 
differences being statistically significant. 
Comments from focus groups help to elucidate both the decreasing perception that 
ordinary people can participate in important decisions and the potential of PDEV II 
activities in core zones to offset that trend. On one hand, respondents in Maradi and Arlit 
in Niger expressed a common sentiment that local politics is too disconnected from 
citizens and based too much on favoritism, so ordinary people are discouraged from 
participating. On the other hand, an imam from Ouahigouya in Burkina Faso noted that 
the PDEV II activities have helped to create a formal consultation network for people in 
the communities, which can help not just in terms of resolving social disputes but also in 
terms of engendering access and confidence for ordinary people to get involved in local 
political matters. It is thus our sense that the scope of PDEV II activities in Burkina 
successfully created the foundations for beneficiaries to engage in local political matters, 
though the activities come amidst a tide of frustration with local politics. 
A second component of social inclusiveness is political participation. To measure this 
dimension, respondents were asked whether or not they have engaged in any of the 
following three activities during the past 12 months: attended a commune/neighborhood 
councilor other public meeting; contacted an elected official; notified the village chief 
about a local problem. Figure 5 graphs the results of an additive index of these three 
items. It suggests minor increases in political participation in Chad and Niger, but a slight 
decrease in Burkina Faso. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
28 
Figure 5: Political Participation 
Table 8: Political Participation 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core .229 .299 .468 X
Baseline Core 
Difference .005 -.001 -.064 X 
Baseline-Midline 
Difference, Non-Core .053 .065 -.108*** X 
Difference in 
Differences, Core versus 
Non-Core
.061 -.007 .057 .030 
* p < .10; ** p < .05; *** p < .01 
The analyses in Table 8 are based on respondents’ average level of reported participation 
across the three items. We find that the P-DEV II program had no discernable impact on 
respondents’ level of political participation, as there are no significant differences in the 
differences in participation rates over time between core and non-core zones between 
baseline and midline. The table suggests a rather low involvement in political activities 
across all countries, and slight increases in core zones in Chad and in the overall pooled 
treatment estimate occurred in statistically indistinguishable ways from changes in non￾core zones over time. While there were initially somewhat higher levels of participation in 
Burkina, this declined between waves. The decline in participation was offset in core 
zones but the difference is not statistically significant. 
One potential explanation is that a selection effect could be at work, whereby those who 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Participates in 0 Participates in 1
Participates in 2 Participates in 3
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
29 
invest fully in the PDEV II activities are the type of people who were already inclined to 
contact officials, engage with chiefs, and attend meetings. If so, this would suggest that 
PDEV II activities can be successful but that they must target behavior changes among 
a wider set of beneficiaries. 
 
Interethnic Marriage (higher values represent increased support) 
A last component of social inclusiveness is ethnic bias. To measure this dimension, 
respondents’ were presented with the following statement: “I tell my children (or I will tell 
my future children) they should only marry people from the same ethnic group as theirs.” 
Respondents could then choose among three options: agree, neither, and disagree. As 
Figure 6 shows, most respondents disagree with this statement in all three countries, and 
the percentage who disagree increased from baseline to midline in all three countries as 
well, suggesting a general increase in support for interethnic marriage (i.e. a decline in 
ethnic bias). 
Indeed, participants in focus groups—such as those in Markoye and Seguenega in 
Burkina Faso—noted that ethnic tensions can sometimes be an issue but that that is 
typically only among youth competing for opportunities or farmers competing for land. 
Beyond those practical concerns, residents of the PDEV zone are moving beyond 
primordial ethnic dislike. 
Figure 6: Support for Interethnic Marriage 
Table 9: Interethnic Marriage 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.411 2.684 2.493 X
Baseline Core Difference -.118 .026 .007 X
Baseline-Midline -.127 .102 .279*** X 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Agree Neither agree nor disagree
Disagree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
30 
Difference, Non-Core 
Difference in Differences, 
Core versus Non-Core .385* .043 -.001 .146 
* p < .10; ** p < .05; *** p < .01 
As Table 9 illustrates, respondents in Chad core zones indicate increased support for 
interethnic marriage, suggesting that P-DEV II interventions had a positive effect there. 
Increases also occur among respondents in Niger and in the overall pooled treatment 
effect estimate but the changes are too small to be statistically significant. In Burkina, 
support for interethnic marriage has increased in non-core and core zones alike, which 
suggests that this positive development was driven by factors other than the P-DEV II 
program. 
Goal 2: Resilience to Violent Extremism
Our analysis of Goal 2: Resilience to Violent Extremism is intended to assess whether 
vulnerable individuals are at risk of becoming radicalized to the point of being willing to 
use violence. We analyze nine separate indicators for Goal 2, distinguishing between: a) 
resilience indicators related to expectations of employment, access to vocational training, 
and political efficacy as factors that enable vulnerable individuals to resist violent 
extremism; and b) vulnerability indicators related to perceptions of and attitudes toward 
violence and extremism as factors that make individuals susceptible to violent extremism. 
Figures 7 through 15 as well as Tables 10 through 15 present the midline results for the 
series of questions related to the concept of resilience. On the nine indicators we analyze, 
we find a mixed pattern of results. On the three indicators of Community Resilience – 
access to jobs, access to vocational training, and political efficacy – we find a general 
decline in these orientations in Chad and Burkina Faso, with positive (or more technically, 
less negative) changes in Burkina P-DEV II core zones compared with non-core zones. 
On the six indicators of Vulnerability – questions related to ethnic and religious community 
divisions, and the justifiability and efficacy of violence – we find generally negative (i.e., 
in the “correct” direction) changes on these perceptions and beliefs in Chad and Burkina 
Faso, with significant differences in differences (i.e., P-DEV II program effects in the 
“correct” direction) on three of the indicators in Burkina. On one indicator, whether the 
respondent disagrees that the US is at war with Islam, we find a program impact in the 
“correct” direction in Chad and in the “wrong” direction in Burkina Faso. The pooled DiD 
treatment effect for the US at War with Islam indicator was significantly negative (i.e. in 
the “correct” direction), the only significant pooled effect seen among these nine 
indicators. 
Resilience Indicators (higher values represent increased resilience) 
To measure respondents’ expectations of employment, they were asked how difficult it is 
to get a job in their respective country today. Figure 7 breaks down their answers by 
country and wave. It shows that expectations of employment have worsened somewhat 
in Chad in Burkina Faso, while there has been only little change in Niger. 
In focus groups—from Markoye, Yatenga, Seguenega, Ouagadougou, Dori, and 
elsewhere—the lack of jobs was a common concern. Frequently, respondents discussing 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
31 
the lack of employment opportunities referred to non-agricultural types of employment, 
which is consistent with conventional wisdom that younger people are less inclined to 
seek work in the fields. Opportunities may exist for PDEV II to encourage employment 
and innovation in domains that complement agriculture (such as value chains), but it 
remains the case that fewer residents of the region consider farm work in their vision of 
future employment. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
32 
Figure 7: Access to Jobs 
Table 10: Access to Jobs 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.959 1.572 1.416 X
Baseline Core 
Difference -.157 -.042 -.023 X 
Baseline-Midline 
Difference, Non-Core -.423*** -.020 -.163** X 
Difference in 
Differences, Core versus 
Non-Core
.164 .051 -.027 .076 
* p < .10; ** p < .05; *** p < .01 
Difference-in-difference analysis shows that the P-DEV II program had no discernable 
effect on respondents’ expectations of employment. The results in Table 10 show 
changes over time in the core zones that are not significantly different from the changes 
in the non-core zones in Chad, Niger and in the pooled treatment effect estimate. Large 
declines between baseline and midline in expectations of employment in Chad and 
Burkina are statistically significant. This decline is especially prominent in Chad, where 
initial expectations of employment during wave 1 were relatively high compared to Niger 
and Burkina. 
A second resilience indicator is Access to Vocational Schools, which was measured by 
presenting respondents’ with the following statement: “Vocational school is accessible for 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
A lot Somewhat
Not at all
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
33 
people like me.” Respondents could choose among three options: disagree, neither, and 
agree. As Figure 8 shows, the portion of respondents who choose “disagree” has 
increased by more than 20% in Chad between waves, suggesting a decreased access to 
vocational school there. For Niger and Burkina Faso, there appears to have been little 
change over time. 
Participants in focus groups expressed two concerns that may offset the effectiveness of 
vocational school promotion. One is the cost of schools in general, which dissuades 
youth and families from considering school an option (see Markoye focus group), and the 
second is the opportunity costs from school attendance, even practical, vocational school. 
Participants in Gorgadji noted that people often prefer to seek something that will provide 
benefits now rather than go through the time it takes to get trained, so vocational schools 
might do better if they were to incorporate in-the-field trainings, internships, and small 
payments for work along with the training. 
Figure 8: Access to Vocational School 
Table 11: Access to Vocational School 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.880 2.000 2.139 X
Baseline Core 
Difference .087 .205 -.214* X 
Baseline-Midline 
Difference, Non-Core -.155 -.085 -.194 X 
Difference in 
Differences, Core 
versus Non-Core
-.289* -.076 .309* -.021 
* p < .10; ** p < .05; *** p < .01 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Disagree Neither agree nor disagree
Agree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
34 
Table 11 reveals a positive effect of P-DEV II activities in Burkina and a negative effect 
of activities in Chad: in Burkina we see improved access to vocational school in the core 
zones relative to the decrease seen in non-core zones, which suggests a positive impact 
of the program in this particular country, while the decrease in access to vocational school 
in Chad has been more pronounced in the core zones, meaning that the P-DEV II program 
seems to have had a negative effect on access to vocational school in Chad. The pooled 
effect of nearly zero reflects the counterbalancing of the positive program impact in 
Burkina and the negative program impact in Chad. 
A final resilience indicator is political efficacy. Respondents were asked to what extent 
they agree with each of the following two statements: a) “My opinions are respected by 
local leaders.” b) “Local government takes into account the opinions of ordinary citizens.” 
They could then choose among three options: not at all, somewhat, and a lot. Figure 9 
graphs the additive index of these two items. Respondents who chose “a lot” for none of 
the two items are classified as having low efficacy, those who chose “a lot” for one item 
as having medium efficacy, and those who chose “a lot” for both as having high efficacy. 
The figure shows a slight increase in efficacy in Chad, no change in Niger, and a rather 
steep decline in Burkina Faso. 
Figure 9: Political Efficacy 
The general trend of discouragement vis-à-vis local government, particularly in Burkina 
Faso, may be a function of the unsettled political context around the time of data 
collection, which came in the aftermath of a popular uprising against former president 
Compaore and as the transitional government faced challenges in the lead up to fresh 
elections. 
Table 12: Political Efficacy 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.835 1.916 2.421 X
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Low efficacy Medium efficacy
High efficacy
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
35 
Baseline Core Difference -.031 -.042 -.338*** X
Baseline-Midline 
Difference, Non-Core .020 -.151 -.400*** X 
Difference in Differences, 
Core versus Non-Core -.067 -.112 .195* -.021 
* p < .10; ** p < .05; *** p < .01 
The analyses in Table 12 are based on respondents’ average level of efficacy across the 
two items. We find an overall decrease in efficacy in P-DEV II core zones in Chad, Niger 
and in our pooled treatment effect estimate, though differences are not statistically 
significant. 
We find a positive program effect in Burkina, where core zones decline significantly less 
from baseline to midline than non-core zones. While respondents in Burkina perceive a 
decrease in efficacy in both core and non-core zones, the difference over time is 
statistically less negative in core as opposed to non-core zones. 
Local political leaders may thus have been preoccupied with their own political 
relationships and less engaged with citizens, but despite the challenges, the effects of 
PDEV II activities in the core zones seemed to have a notable effect. 
Vulnerability Indicators (higher values represent increased vulnerability) 
Unlike other indicators in this report, the vulnerability indicators are coded in such way 
that lower values are desirable, as higher values represent increased vulnerability. This 
means that normatively desirable P-DEV II program effects should be reflected in 
negative coefficients in the following tables. 
The first vulnerability indicator is perception of ethnic differences, which is measured by 
asking respondents to what extent they feel that ethnic differences tend to divide people 
in their village/neighborhood. Figure 10 illustrates that the percentage of respondents who 
answer “not at all” has increased in Chad and Burkina; Niger has experienced the 
opposite trend, with more respondents choosing the option “a lot.” 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
36 
Figure 10: Perceived Ethnic Differences 
Table 13: Perceived Ethnic Differences 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.494 1.409 1.316 X
Baseline Core 
Difference -.003 .012 .093 X 
Baseline-Midline 
Difference, Non-Core -.214*** .184** -.043 X 
Difference in 
Differences, Core 
versus Non-Core
.014 -.003 -.172** -.048 
* p < .10; ** p < .05; *** p < .01 
Analysis in Table 13 reveals a positive program effect in Burkina. A decline in vulnerability 
occurs across all zones in Chad, while core zone in Niger and in the pooled treatment 
estimate reveal statistically insignificant differences. In Burkina, positive impact in P-DEV 
II core zones is shown through a significant decrease in perceived ethnic differences 
among respondents. Consistent with the focus group commentary noted above, the 
general trend seems to be away from vulnerability due to ethnic divisions, even if some 
ethnic tensions persist regarding practical issues like land and jobs. 
In Chad, there was a significant decrease in perceived differences between baseline and 
midline across all zones, whereas Niger there was a general increase in perceived 
differences. Changes to core zones in Niger and in the pooled treatment estimate 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Not at all Somewhat
A lot
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
37 
Respondents were also asked about religious differences, i.e., the extent they feel that 
religious differences tend to divide people in their village/neighborhood. Figure 11 breaks 
down their responses by country and wave. Similar to the ethnic differences indicator, the 
figure suggests improvements over time in Chad and Burkina Faso, with more 
respondents answering “not at all.” In Niger, again, there seems to be the opposite 
happening, with a greater percentage of respondents choosing the option “a lot.” 
One potential explanation for the rise in religious-based vulnerability in Niger is the 
struggle for control of religious narratives as extremist attacks became more common 
there. In the focus group discussion in Arlit, participants noted that leaders sometimes 
try to mastermind people’s attitudes regarding social tensions, so competing narratives—
both in Hausa-Tuareg ethnic terms and also in moderate-extremist religious terms—could 
contribute to an increased sense of vulnerability in Niger. 
Figure 11: Perceived Religious Differences 
Table 14: Perceived Religious Differences 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.432 1.472 1.311 X
Baseline Core 
Difference .006 .036 .129* X 
Baseline-Midline 
Difference, Non-Core -.204*** .155 -.066 X 
Difference in 
Differences, Core versus 
Non-Core
.028 .006 -.163* -.037 
* p < .10; ** p < .05; *** p < .01 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Not at all Somewhat
A lot
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
38 
The perceived religious differences results show exactly the same pattern in Table 14 as 
was seen for ethnic differences in Table 13: a positive impact of P-DEV II in Burkina, 
where the decrease in perceived religious differences is concentrated exclusively in the 
core zones. In Chad, there has been an overall decline in perceived religious differences 
among respondents across core and non-core zones, while in Niger there was more 
(though in this case statistically insignificant) perceived religious difference. Decreases in 
perceived differences in the pooled treatment effect estimate are statistically insignificant. 
Another vulnerability indicator is the belief in the justifiability of religious violence. To 
measure this dimension, respondents were asked how often they feel that using arms 
and violence against civilians in defense of one’s religion is justified. Respondents could 
then choose among three options: never, sometimes, and often. Figure 12 breaks down 
their responses by country and wave. In the baseline survey, a large majority of 
respondents in all three countries (> 70%) chooses the option “never.” Over time, the 
portion of respondents who think that religious violence is never justified has become 
even greater, especially in Burkina Faso. 
Figure 12: Religious Violence Is Justified 
Table 15: Religious Violence Is Justified 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.449 1.266 1.199 X
Baseline Core Difference -.059 .072 .125* X
Baseline-Midline 
Difference, Non-Core -.222** .023 -.126*** X 
Difference in Differences, 
Core versus Non-Core -.003 -.134 -.131* -.083 
* p < .10; ** p < .05; *** p < .01 
Table 15 illustrates a positive program effect for Burkina, where core zone respondents 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Never Sometimes
Often
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
39 
find religious violence to be less justifiable than at baseline. In Chad a general decrease 
in the perceptions that religious violence is justified across both core and non-core zones. 
In Niger, there is a suggestion that respondent perceptions of the justifiability of religious 
violence declined in core zones more sharply than in non-core zones, though this 
difference is not statistically significant. Decreases in the justifiability of violence in the 
pooled treatment effect estimate are statistically insignificant, though, from a qualitative 
standpoint, focus group participants quite frequently noted that Islam is a religion of peace 
and that forgiveness and tolerance are important components of the religion (see, for 
example, FGDs in Tougu and Seguenega). 
Respondents were also asked how often they think violence is an effective method to 
solve problems. Again, they could choose among three options: never, sometimes, and 
often. Figure 13 reveals that a majority of respondents in each country (> 65%) answers 
this question with “never.” All three countries exhibit a positive development regarding 
this particular indicator, with the percentage of respondents choosing “never” becoming 
greater in Chad, Niger, and Burkina between the baseline and midline. 
Figure 13: Violence Is Effective to Solve Problems 
Table 16: Violence Is Effective to Solve Problems 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.470 1.308 1.352 X
Baseline Core 
Difference -.074 .076 .179** X 
Baseline-Midline 
Difference, Non-Core -.272*** .003 -.151*** X 
Difference in 
Differences, Core 
versus Non-Core
.055 -.172 -.047 -.050 
* p < .10; ** p < .05; *** p < .01 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Never Sometimes
Often
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
40 
Table 16 shows no effect for P-DEV II programming. There has been a significant 
decrease in the perception that violence is an effective method to solve problems from 
baseline to midline in both Chad and Burkina in both core and non-core zones and can 
therefore not be attributed to the P-DEV II program. In Niger and in our pooled treatment 
effect estimate, respondents’ perception of effectiveness decreases in core zones, but 
these differences do not occur in statistically distinguishable ways from the non-core 
zones. According to interviews with leaders in Agadez, Arlit, and elsewhere, civil society 
groups, women’s groups, and other actors have worked hard to stigmatize the use of 
violence. As a result, PDEV II activities in core areas, while likely contributing in a positive 
sense, do not distinguish those areas from non-core zones where other programs and 
activities aim to address the same issue. 
Next, we asked respondents whether they disagree or agree with the statement that 
“violence in the name of Islam can be justified.” As with the last two questions, Figure 14 
reveals that a majority of respondents in all three countries (> 60%) rejects violence. The 
portion of respondents who choose the option “disagree” has become greater over time 
in all three countries. 
Figure 14: Violence in the Name of Islam 
Table 17: Violence in the Name of Islam 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.539 1.574 1.540 X
Baseline Core 
Difference -.037 .023 -.099 X 
Baseline-Midline 
Difference, Non-Core -.461*** -.395*** -.411*** X 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Disagree Neither agree nor disagree
Agree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
41 
Difference in 
Differences, Core versus 
Non-Core
.050 -.028 .143 .049 
* p < .10; ** p < .05; *** p < .01 
Table 17 shows no effect for P-DEV II programming. In all three countries, there has been 
a large and significant decrease in the perception that violence in the name of Islam can 
be justified from baseline to midline. This decrease has been happening in core and non￾core zones alike and can therefore not be attributed to the P-DEV II program. Among 
respondents in core zones, slight increases in justifiability of violence in the name of Islam 
occur in Chad, Burkina, and in our pooled treatment effect estimate but these differences 
are statistically insignificant. Interviews again confirm that numerous actors and leaders 
across the region have been working to combat an association between violence and 
Islam, making it difficult to isolate an effect from the PDEV II activities. 
A last vulnerability indicator is anti-West attitudes. This dimension is measured by asking 
respondents whether they disagree or agree with the following statement: “The United 
States is at war against Islam, not terrorism.” Figure 15 breaks down their responses by 
country and wave. In wave 1, a majority of respondents in Chad and Burkina disagreed 
with the statement, whereas the majority of surveyed individuals in Niger chose to agree 
with it. In all three countries, the percentage of respondents disagreeing with the notion 
that the U.S. is at war with Islam has increased over time, suggesting decreasing anti￾West attitudes in Chad, Niger, and Burkina Faso. 
In Niger, the focus group in Balleyara noted that Niger has been Muslim for a long time 
and the country has had quite good relations with the United States, so it would be 
shortsighted to think that the US is fighting Islam rather than the current wave of terrorism. 
However, respondents in Arlit and elsewhere noted that the US is also engaged in conflict 
with a number of other Muslim countries, which raises concerns for some. More 
generally, the overwhelming predominance of Islam in Niger compared to Burkina Faso 
and Chad, as well as the greater degree to which religion shapes sociopolitical life in 
Niger coupled with the constructed perception US-Muslim tensions during the period of 
the War on Terror, perhaps explains the greater tendency to suspect US aggression 
toward Islam there. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
42 
Figure 15: U.S. Is at War with Islam 
Table 18: U.S. Is at War with Islam 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.398 2.061 1.659 X
Baseline Core 
Difference .057 .244 -.036 X 
Baseline-Midline 
Difference, Non-Core .019 -.037 -.440*** X 
Difference in 
Differences, Core versus 
Non-Core
-.286** -.712** .266** -.242*
* p < .10; ** p < .05; *** p < .01 
Table 18 shows a positive P-DEV II program effect in Chad, Niger, and in our pooled 
treatment effect estimate. In Chad and Niger, the decrease in anti-West attitudes is 
concentrated in the core zones, thus suggesting a positive effect of the P-DEV II program. 
However, in Burkina, the core zones exhibit a significant increase in anti-West attitudes 
relative to the decrease in anti-West attitudes found in non-core zones. 
PDEV II activities seem to have made an important impact, however, largely through 
cooperation with imams (according to interviews). The apparent negative impact of PDEV 
II activities in Burkina Faso may again be a function of the political climate: because the 
US had been a strong supporter of former President Blaise Compaore, many came to 
view the US with suspicion, and this may have influenced their reaction to the US more 
broadly. 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Disagree Neither agree nor disagree
Agree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
43 
Goal 3: Youth Outlook
Figures 16 through 20 as well as Tables 19 through 23 show the results for questions 
related to Goal Level Index 3: Youth Outlook, defined as the individual and collective 
vision people have about young people in their communities. This concept is 
operationalized using three categories of indicators: a) economic outlook which evaluates 
attitudes toward and practical vision of future careers and economic potential; b) 
educational outlook which evaluates expectations regarding education and the learning 
environment; and c) civic outlook which comprises engagement in politics, civil society, 
and local decision-making. The tables in this section present the results for the full sample 
in each country, that is, individuals of all ages; where relevant we discuss differences in 
the results for the sub-sample of youth in the analysis, i.e., individuals between 18-30 
years of age. We caution that this sub-sample comprises less than 50% of the total, or 
slightly more than 1000 respondents in the midline wave spread across three countries, 
so the results are less robust than those for the full sample. 
Our analysis of Goal 3 Youth Outlook indicators finds a mixed pattern of results for the 
five indicators. On general life satisfaction and economic perceptions, average responses 
were worse in Chad though no different in core versus non-core zones. Economic 
perceptions were significantly improved in Burkina core versus non-core zones. 
Educational inclusiveness improved in all three countries, with significantly greater 
improvements in core versus non-core zones in Chad, and political interest and 
knowledge increased in all three countries, with greater differences in differences on 
knowledge in both Niger and Burkina Faso. The pooled DiD treatment effect across all 
countries was significant only for political knowledge among this set of indicators. 
Life Satisfaction and Economic Outlook (higher values represent improved outlook) 
To measure life satisfaction, respondents were presented with a card showing a ladder 
representing the “ladder of life.” The ladder consists of eleven steps, ranging from 0 to 
10. Respondents were then asked the following question: “Let’s suppose the top of the 
ladder is the best possible life for you; and the bottom, the worst possible life for you. On 
which step of the ladder do you personally stand at the present time?” Figure 16 groups 
levels 0 through 3 into the category “low satisfaction,” levels 4 through 6 into “medium 
satisfaction,” and 7 through 10 into “high satisfaction.” 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
44 
Figure 16: Life Satisfaction 
Table 19: Life Satisfaction 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 4.624 3.842 4.250 X
Baseline Core 
Difference .038 .697*** -.013 X 
Baseline-Midline 
Difference, Non-Core -.601** .886*** .422 X 
Difference in 
Differences, Core 
versus Non-Core
-.004 -.601 .404 -.090 
* p < .10; ** p < .05; *** p < .01 
The analyses in Table 19 a based on the full range of eleven points, and show no effect 
for P-DEV II programming. Among respondents, life satisfaction has decreased over time 
in Chad but increased in Niger. In both cases, these seem to be general developments 
that are independent of the P-DEV II program. A comparison of core and non-core 
differences over time reveals decreases in life satisfaction in core zones in Chad, Niger, 
and in our pooled treatment effect analysis that are statistically insignificant. 
To measure perceptions of the economy, respondents were asked whether the economy 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Low satisfaction Medium satisfaction
High satisfaction
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
45 
of their country is worse, about the same, or better than it was a year ago. Figure 17 
breaks down their responses by country and wave. It shows that in Niger, perceptions of 
the economy have remained quite stable over time, whereas in Burkina and especially in 
Chad they have become somewhat more negative. In the midline survey, more than 40% 
of Chadian said their economy is worse than it was a year ago; in the baseline survey, 
that figure was only about 25%. The decline in economic outlook in Chad likely reflects 
the mismanagement of oil revenues that resulted in a broad economic crisis, while 
respondent’s economic outlook in Burkina Faso from 2013 to 2015 was almost certainly 
impacted by the uncertainty over a new political regime. 
Figure 17: Economic Outlook 
Table 20: Economic Outlook 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.080 2.191 2.373 X
Baseline Core 
Difference .023 .061 -.416*** X 
Baseline-Midline 
Difference, Non-Core -.358** .011 -.293*** X 
Difference in 
Differences, Core versus 
Non-Core
-.016 -.125 .392* .059 
* p < .10; ** p < .05; *** p < .01 
Table 20 illustrates a P-DEV II program effect in Burkina Faso, where respondents in core 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
worse the same
better
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
46 
zones report an improved economic outlook. In Burkina, perceptions have worsened in 
the non-core zones but there is a significant positive difference in difference in the core 
zones, suggesting a positive P-DEV II program effect. In Chad, non-core and core zone 
respondent perceptions of the economy have significantly worsened over time; this 
negative development can therefore not be attributed to the P-DEV II program. 
Improvements in respondent economic outlook in the pooled treatment effect estimate 
are statistically insignificant. 
Educational Outlook (higher values represent improved outlook) 
The second category of indicators with Goal 3 is respondents’ educational outlook. 
Specifically, we measure respondents’ attitudes regarding educational inclusiveness by 
asking whether they agree or disagree with the following statement: “Middle school is only 
for boys, not girls.” Figure 18 shows that the vast majority of respondents disagree with 
this statement. Over time, respondents’ attitudes regarding education have become even 
more inclusive. 
Figure 18: Gender Inclusiveness of Middle School 
Table 21: Gender Inclusiveness of Middle School 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.644 2.601 2.615 X
Baseline Core 
Difference -.084 -.002 .087 X 
Baseline-Midline 
Difference, Non-Core -.049 .184** .294*** X 
Difference in 
Differences, Core versus .218* -.071 -.113 .007 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
Agree Neither agree nor disagree
Disagree
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
47 
Non-Core
* p < .10; ** p < .05; *** p < .01 
Table 21 provides additional evidence that in all three countries, attitudes regarding 
education have become more inclusive over time. In Chad, this positive development can 
be attributed to a P-DEV II program effect: improvements in attitudes toward 
inclusiveness are concentrated in the core zones. Small improvements in attitudes toward 
gender inclusiveness in in middle school in in the pooled treatment effect estimate are 
statistically insignificant. Given efforts to improve educational access across the region 
by numerous donors, international organizations, and civil society actors, it should come 
as little surprise that the positive effects of PDEV II in Niger and Burkina Faso are masked 
by improvements in non-core zones, and the PDEV impact in Chad is particularly 
noteworthy. 
Civic Outlook (higher values represent improved outlook) 
The last category of indicators we examine in this report is civic outlook. Within this 
category, we first look at interest in community affairs. This dimension is measured by 
asking respondents whether they have very little interest, some interest, or a great deal 
of interest in local community affairs. As Figure 19 shows, the percentage of respondents 
who say they have “very little interest” in community affairs has decreased over time, 
whereas the percentage of respondents who choose the option “a great deal of interest” 
has increased in all three countries. 
This trend is interesting to note when taken together with the finding that fewer people 
feel empowered in terms of political decision making. In in-depth interviews, (for example 
in Ouahigouya and Seytenga in Burkina Faso), CSO leaders and imams expressed that 
people—youth in particular—are becoming more informed and that community efforts to 
establish of communication really help to reinforce both knowledge and concern over 
political affairs. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
48 
Figure 19: Interest in Community Affairs 
Table 22: Interest in Community Affairs 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.577 1.651 2.196 X
Baseline Core 
Difference .064 -.099 -.291* X 
Baseline-Midline 
Difference, Non-Core .206*** .638*** .265* X 
Difference in 
Differences, Core 
versus Non-Core
-.018 -.003 .220 .032 
* p < .10; ** p < .05; *** p < .01 
Table 22 provides additional evidence for the observation that there has been an increase 
in political interest in all three countries. Positive changes are registered in all three 
countries in both non-core and core zones, and do not occur between core and non-core 
zones so these changes cannot be attributed to the P-DEV II program. 
The second component of civic outlook is political knowledge. To measure this dimension, 
respondents were asked two questions: a) “Do you know how long the term of office is 
for the President?” b) “Do you know how many seats there are in the National Assembly?” 
Figure 20 shows the percentages of respondents who gave none, one, and two correct 
answers to these questions, broken down by country and wave. Political knowledge 
appears to be highest in Niger, followed by Burkina, then Chad. In all three countries, 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
very little interest some interest
a great deal of interest
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
49 
political knowledge seems to have increased between waves. Again, these findings are 
consistent with qualitative reports that people in the region, and especially youth, are 
gaining improved access to information. 
Figure 20: Political Knowledge 
Table 23: Political Knowledge 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core .229 .479 .331 X
Baseline Core 
Difference .107*** -.048 .018 X 
Baseline-Midline 
Difference, Non-Core .094 -.032 .012 X 
Difference in 
Differences, Core 
versus Non-Core
-.026 .175** .079** .067* 
* p < .10; ** p < .05; *** p < .01 
Table 23 illustrates a positive effect of P-DEV II programming in Niger, Burkina, and in 
our pooled treatment effect estimate, as positive and significant increases occur in 
respondent political knowledge from the baseline to midline. 
Examining P-DEV Effects with Youth Subgroup Analysis 
Finally, we conducted another series of analyses on each of the five indicators in this 
section, limiting the sample only to youth, or people 30 years of age or younger. The 
results showed several anomalous results, in that core zone youth in Chad declined 
dramatically more on life satisfaction than non-core zone youth in Chad, and core zone 
0 20 40 60 80 100 percent
Chad Niger Burkina
Wave 1 Wave 2 Wave 1 Wave 2 Wave 1 Wave 2
0 correct answers 1 correct answer
2 correct answers
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
50 
youth in Burkina declined more on Gender Inclusiveness of Middle Schools than non-core 
zone Burkina youth. Otherwise the same pattern is seen as in the full sample, i.e. mainly 
insignificant, or on occasion, weakly positive DiD program impacts. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
51 
C. The Impact of “Low Intensity” Versus “High Intensity” P-DEV II Programming 
In this section, we present the impact of P-DEV II programming by dividing zones into 
those in which no activities were implemented, those in which a relatively small number 
of activities were carried out, and those in which a comparatively high number of activities 
were implemented. It may be, for example, that significant program effects may be found 
by examining those zones where more intensive programming took place, and that the 
null or weak effects found in the previous section may be masking important impacts for 
the high intensity core zones of the program. For this distinction, we rely on information 
provided by the P-DEV II activity tracker and discussed in Section II-D and III-A above. 
Non-core zones are classified as zones in which no activities were implemented. Core 
zones that experienced below the median number of activities per zone for each country 
were classified as “low treatment,” and those higher than the median on total activities 
were classified as “high treatment.” For Chad, the median number of activities per zone 
is 22, for Niger it is 33, and for Burkina Faso it is 65.4
Similar to the models estimated in Section III-B above, these effects for “low intensity” 
and “high intensity” zones are calculated via a DiD approach: we regress the 19 goal￾level indicators on dichotomous variables for no/low/high treatment zones, wave 1/2, and 
country, as well as the interactions between these variables. If the program had the 
intended effects, we would see an improvement (a “difference in differences”) in the low 
treatment zones relative to the non-core zones, and we would see an even greater 
improvement (or DiD) in the high treatment zones from baseline to midline relative to the 
changes in the non-core zones between waves. 
In equation form, these DiD models may be expressed as: 
(3) Yit = a + b1LowCore
i + b 2HighCore
i + b 3Wave
t + b 4LowCore
i *Wave
t + b 5HighCore
i *Wave
te
it
where Y (a given outcome) at a given point in time is equal to: a common intercept or 
starting point (α), an effect (β1) of whether the individual is in a low intensity core zone, 
an effect (β2) of whether the individual is in a high intensity core zone, an effect (β3) of a 
given wave of observation (baseline versus midline) on all individuals, an interaction effect 
(β4) of wave of observation with low intensity core zone status, an interaction effect (β5) 
of wave of observation with high intensity core zone status, and an idiosyncratic error 
term (εit). Following the DiD logic explained above, β4 represents the estimated impact of 
being in a low intensity core zone relative to a non-core zone, and β5 represents the 
estimating impact of being in a high intensity core zone, relative to a non-core zone. If 
there are significant program effects concentrated only in high (or in low) intensity zones, 
we would then expect a significant regression coefficient only for β5 (or β4). 
The tables are arranged as follows: The first row illustrates the average value for the 
respective indicator in each country’s non-core zones during the first wave of the survey, 
 
4
 In addition, we created a classification into no/low/high treatment zones for each of the twelve types of activities shown 
in Section III-A of this report. We then regressed each indicator on the type of activity it is most likely to be influenced 
by, rather than on the additive index of all activities. For example, the indicator “religious violence is justified” is 
regressed on activity type “2.3 increased dialogue with religious leaders.” This did not meaningfully alter the majority of 
our findings reported in this section; therefore, these additional results are not included in this report. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
52 
i.e. before the implementation of the P-DEV II program (α). The second and third rows 
show the average difference between non-core and low (β1) treatment zones, and then 
between non-core and high (β2) treatment zones at baseline, and whether these 
difference is statistically significant. In the fourth row, we present the average difference 
between non-core zones at midline and baseline (β3) and whether this difference is 
statistically significant. 
The fifth and sixth rows are the ones in which the actual effects of the P-DEV II program 
are shown. In the fifth row, we take the difference between midline and baseline in the 
low treatment zones and subtract the difference between midline and baseline in the non￾core zones – this difference in difference corresponds to β4 in equation (3). In the sixth 
row, we take the difference between midline and baseline in the high treatment core zones 
and subtract the difference between midline and baseline in the non-core zones – this 
difference corresponds to β5 in equation (3). In addition to the effects for each country, 
the fifth and sixth rows also contain a pooled treatment effect, which shows whether the 
program had an effect across all three countries in the given type of zone (low versus 
high treatment). For the reader’s convenience, the fifth and sixth rows are color-coded: 
coefficients representing a desired effect of the program are shown in green, whereas 
coefficients representing a detrimental effect of the program are shown in red. Again, 
pooled models are estimated first, followed by variants that allow country-specific 
treatment effects. 
Goal 1: Social Cohesion
Tables 24 through 28 present the midline results for the series of questions related to 
Goal 1: Social Cohesion. Social cohesion is a broad concept that is captured through two 
separate indicators: 1) interpersonal and institutional trust; and 2) social inclusiveness in 
the community. 
Interpersonal and Institutional Trust (higher values represent increased trust) 
To measure interpersonal trust, respondents were asked whether they disagree or agree 
that “most people are willing to help if you ask for help.” As Table 24 shows, in the zones 
in which a relatively low number of activities were implemented, there are no effects of 
the P-DEV II program on how respondents answer this question. In the high treatment 
zones in Burkina, interpersonal trust has increased over time, while in both Chad and 
Niger, it has actually decreased between waves. These results are somewhat different 
from those in Section B, which suggest that the P-DEV II program had no effect on 
interpersonal trust. While the pooled effect is still irrelevant in these models, there appear 
to be offsetting positive and negative impacts in the high intensity zones across the three 
countries. There is no evident reason for this anomalous pattern of effects. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
53 
Table 24: Interpersonal Trust 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.295 2.298 2.616 X
Baseline Low Core 
versus Non-Core .008 .033 -.130* X 
Baseline High Core 
versus Non-Core .105 .011 -.390*** X 
Baseline-Midline 
Difference, Non-Core -.016 .136 -.308*** X 
Difference in 
Differences, Low Core 
versus Non-Core
.157 -.086 .086 .028 
Difference in 
Differences, High Core 
versus Non-Core
-.273*** -.223* .284*** -.119 
* p < .10; ** p < .05; *** p < .01 
Institutional trust is measured through five individual trust items. Each respondent was 
presented with five separate statements: a) I trust local authorities; b) I trust the central 
government; c) I trust religious leaders; d) I trust NGOs; e) I trust the police. For each of 
these statements, they could choose among three options: disagree, neither, or agree. 
The analyses in Table 25 are based on respondents’ average level of trust across the five 
institutions. It shows that there has been little change in institutional trust between wave 
1 and wave 2, neither in the non-core nor in the low or high treatment zones. These results 
are similar to those in Section B, which suggest that the P-DEV II program has no effect 
on institutional trust. For both interpersonal and institutional trust, there are no pooled DiD 
effects, neither in the low treatment nor in the high treatment zones. 
Table 25: Institutional Trust 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.487 2.589 2.737 X
Baseline Low Core versus 
Non-Core -.125 -.046 -.192** X 
Baseline High Core 
versus Non-Core -.189** -.240*** -.347*** X 
Baseline-Midline 
Difference, Non-Core -.008 .009 -.053 X 
Difference in Differences, 
Low Core versus Non￾Core
.140 .002 .059 .055 
Difference in Differences, 
High Core versus Non-
-.158 .099 .079 -.006 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
54 
Core
* p < .10; ** p < .05; *** p < .01 
Social Inclusiveness (higher values represent increased inclusiveness) 
To measure community decision-making, respondents were asked to what extent 
ordinary people from the commune/neighborhood participate when important decisions 
are being made. Table 26 shows that in Burkina, there has been a significant increase in 
this measure of social inclusiveness in the low treatment zones relative to the no 
treatment zones. These results are similar to those in Section B, which suggest that the 
P-DEV II program in Burkina had a positive impact on respondents’ perceptions of being 
included when decisions are made in their community. However, the high intensity zone 
effect, while similar in magnitude to the low intensity effect, does not reach the level of 
statistical significance. 
Table 26: Community Decision-making 
Chad Niger Burkina Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.947 2.190 2.518 X
Baseline Low Core 
versus Non-Core .039 -.049 -.378*** X 
Baseline High Core 
versus Non-Core .034 -.012 -.447*** X 
Baseline-Midline 
Difference, Non-Core -.102 -.170 -.404*** X 
Difference in 
Differences, Low 
Core versus Non￾Core
-.021 -.044 .295*** .050 
Difference in 
Differences, High 
Core versus Non￾Core
-.336* -.168 .366 -.087 
* p < .10; ** p < .05; *** p < .01 
To measure political participation, respondents were asked whether or not they have 
engaged in any of the following three activities during the past 12 months: attended a 
commune/neighborhood councilor other public meeting; contacted an elected official; 
notified the village chief about a local problem. The analyses in Table 27 are based on 
respondents’ average level of participation across the three items. The table illustrates 
that in Burkina, there has been a significant increase in participation in the high treatment 
zones relative to the no treatment zones. These results are different from those in Section 
B, which suggest that the P-DEV II program has no effect on participation. When isolating 
high intensity treatment zones, an effect on participation relative to non-core zones 
becomes apparent in the Burkina context. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
55 
Table 27: Political Participation 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core .229 .299 .468 X
Baseline Low Core versus 
Non-Core .009 .039 -.059 X 
Baseline High Core 
versus Non-Core -.001 -.068 -.073*** X 
Baseline-Midline 
Difference, Non-Core .053 .065 -.108*** X 
Difference in Differences, 
Low Core versus Non￾Core
.122 -.040 .042 .025 
Difference in Differences, 
High Core versus Non￾Core
.009 .048 .085* .038 
* p < .10; ** p < .05; *** p < .01 
To measure ethnic bias, respondents’ were presented with the following statement: “I tell 
my children (or I will tell my future children) they should only marry people from the same 
ethnic group as theirs.” Respondents could then choose among three options: agree, 
neither, and disagree. As Table 28 illustrates, respondents have become significantly 
more tolerant of interethnic marriage in the low treatment zones in Chad and in the high 
treatment zones in Burkina. These results are again different from those in Section B, 
which suggest that the P-DEV II program has no effect on ethnic bias. It is not apparent, 
however, why the Chad effect would be seen only in the low intensity zones, nor why the 
overall pooled effect would be significant only for low intensity zones. 
Table 28: Support for Interethnic Marriage 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.411 2.684 2.493 X
Baseline Low Core 
versus Non-Core -.187* -.022 .050 X 
Baseline High Core 
versus Non-Core .253*** .215** -.117** X 
Baseline-Midline 
Difference, Non-Core -.127 .102 .279*** X 
Difference in 
Differences, Low Core 
versus Non-Core
.477** .094 -.060 .179*
Difference in 
Differences, High Core 
versus Non-Core
-.006 -.151 .153*** .001 
* p < .10; ** p < .05; *** p < .01 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
56 
Goal 2: Resilience to Violent Extremism
Tables 29 through 35 present the midline results for the series of questions related to 
Goal Level Index 2: Resilience to Violent Extremism. This concept is intended to assess 
whether vulnerable individuals are at risk of becoming radicalized to the point of being 
willing to use violence. We distinguish between: a) resilience indicators related to 
expectations of employment, access to vocational training, and political efficacy as factors 
that enable vulnerable individuals to resist violent extremism; and b) vulnerability 
indicators related to perceptions of and attitudes toward violence and extremism as 
factors that make individuals susceptible to violent extremism. 
Resilience Indicators (higher values represent increased resilience) 
To measure respondents’ expectations of employment, they were asked how difficult it is 
to get a job in their respective country today. Possible responses were “a lot,” “somewhat,” 
and “not at all.” Table 29 shows that for high treatment zones, expectations of employment 
have significantly improved in Chad but have worsened in Burkina Faso. These results 
are different from those in Section B, which suggest that the P-DEV II program has no 
effect in any of the three country contexts on expectations of employment. Instead, there 
is a positive impact in one country, Chad, but an anomalous negative impact in another, 
Burkina. In addition, there is a positive pooled treatment effect in the high treatment 
zones, suggesting that the P-DEV II program has improved expectations of employment 
across the three countries. This pooled effect, however, is mostly driven by the positive 
development in Chad. 
Table 29: Access to Jobs 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect 
Baseline, Non-Core 1.959 1.572 1.416 X 
Baseline Low Core versus 
Non-Core -.031 .030 -.072 X 
Baseline High Core 
versus Non-Core -.322*** -.161* .116* X 
Baseline-Midline 
Difference, Non-Core -.423*** -.020 -.163** X 
Difference in Differences, 
Low Core versus Non￾Core 
.047 .024 .008 .015 
Difference in Differences, 
High Core versus Non￾Core 
.319** .096 -.141** .177** 
* p < .10; ** p < .05; *** p < .01 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
57 
Access to vocational school was measured by presenting respondents’ with the following 
statement: “Vocational school is accessible for people like me.” Respondents could then 
choose among three options: disagree, neither, and agree. As Table 30 illustrates, there 
has been a significant improvement in access to vocational school in low treatment zones 
in Burkina but a decrease in access in high treatment zones in Chad. These results are 
similar to those in Section B, which suggest that the P-DEV II program has a positive 
effect on access to vocational school in Burkina but a negative effect in Chad, though it 
is unclear why the impact would be concentrated among low intensity zones in Burkina. 
Table 30: Access to Vocational School 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect 
Baseline, Non-Core 1.880 2.000 2.139 X 
Baseline Low Core 
versus Non-Core .088 .278* -.217 X 
Baseline High Core 
versus Non-Core .087 .082 -.209 X 
Baseline-Midline 
Difference, Non-Core -.155 -.085 -.194 X 
Difference in 
Differences, Low Core 
versus Non-Core 
-.147 -.257 .357* -.037 
Difference in 
Differences, High Core 
versus Non-Core 
-.410** .225 .224 .004 
* p < .10; ** p < .05; *** p < .01 
To measure political efficacy, respondents were asked to what extent they agree with 
each of the following two statements: a) “My opinions are respected by local leaders.” b) 
“Local government takes into account the opinions of ordinary citizens.” They could then 
choose among three options: not at all, somewhat, and a lot. The analyses in Table 31 
are based on respondents’ average level of efficacy across the two items. The results 
shown are similar to those in Section B, which suggest that the P-DEV II program has a 
positive effect on political efficacy in Burkina. The difference between the impacts in low 
and high intensity zones in Burkina is extremely small, which is consistent with the finding 
earlier that core zones in general experienced a significant positive change in efficacy. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
58 
Table 31: Political Efficacy 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.835 1.916 2.421 X
Baseline Low Core versus 
Non-Core .085 .042 -.320** X 
Baseline High Core versus 
Non-Core -.046 -.180* -.369*** X 
Baseline-Midline Difference, 
Non-Core .020 -.151 -.400*** X 
Difference in Differences, 
Low Core versus Non-Core .086 -.142 .198* .004 
Difference in Differences, 
High Core versus Non-Core -.179 -.067 .187 -.051 
* p < .10; ** p < .05; *** p < .01 
Vulnerability Indicators (higher values represent increased vulnerability) 
Unlike other categories in this report, the vulnerability indicators are coded in such way 
that lower values are desirable, as higher values represent increased vulnerability. The 
first component of these indicators is ethnic differences, which is measured by asking 
respondents to what extent they feel that ethnic differences tend to divide people in their 
village/neighborhood. Respondents could then choose among three options: not at all, 
somewhat, and a lot. Table 32 shows that there has been a significant decrease in ethnic 
differences in the high treatment zones in Burkina over time. These results are consistent 
with those in Section B, which suggest that P-DEV II has an ameliorating effect on 
perceptions of ethnic differences in Burkina. 
Table 32: Ethnic Differences 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.494 1.409 1.316 X
Baseline Low Core versus 
Non-Core -.130 -.005 .035 X 
Baseline High Core versus 
Non-Core .166 .041 .195** X 
Baseline-Midline 
Difference, Non-Core -.214*** .184** -.043 X 
Difference in Differences, 
Low Core versus Non￾Core
.068 .011 -.167 -.021 
Difference in Differences, 
High Core versus Non￾Core
-.090 -.028 -.178*** -.095 
* p < .10; ** p < .05; *** p < .01 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
59 
To measure a second vulnerability indicator, religious differences, respondents were 
asked to what extent they feel that religious differences tend to divide people in their 
village/neighborhood. Again, respondents could choose among three options: not at all, 
somewhat, and a lot. As Table 33 illustrates, there has been a significant decrease in 
religious differences in the high treatment zones in Burkina over time. These results are 
similar to those in Section B, which suggest that the P-DEV II program has an ameliorating 
effect on religious differences in Burkina. 
Table 33: Religious Differences 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.432 1.472 1.311 X
Baseline Low Core versus 
Non-Core -.055 .030 .045 X 
Baseline High Core 
versus Non-Core .087 .047 .273*** X 
Baseline-Midline 
Difference, Non-Core -.204*** .155 -.066 X 
Difference in Differences, 
Low Core versus Non￾Core
-.002 .012 -.128 -.032 
Difference in Differences, 
High Core versus Non￾Core
.031 -.004 -.220** -.048 
* p < .10; ** p < .05; *** p < .01 
To measure the justifiability of religious violence, respondents were asked how often they 
feel that that using arms and violence against civilians in defense of one’s religion if 
justified. Respondents could then choose among three options: never, sometimes, and 
often. Table 34 shows that there has been a significant decrease in the perceptions that 
religious violence is justified in the high treatment zones in Burkina. These results are 
consistent with those in Section B, which suggest that the P-DEV II program has an 
ameliorating effect on the justifiability of religious violence in Burkina. There is also a 
significant pooled treatment effect for low intensity zones, reflecting the relatively 
consistent negative impact across the three country contexts. 
Table 34: Religious Violence Is Justified 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.449 1.266 1.199 X
Baseline Low Core versus 
Non-Core .037 .123 .040 X 
Baseline High Core versus 
Non-Core -.184* -.011 .264*** X 
Baseline-Midline 
Difference, Non-Core -.222** .023 -.126*** X 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
60 
Difference in Differences, 
Low Core versus Non-Core -.133 -.209 -.041 -.133*
Difference in Differences, 
High Core versus Non￾Core
.153 -.010 -.281*** -.009 
* p < .10; ** p < .05; *** p < .01 
Respondents were also asked how often they think violence is an effective method to 
solve problems. Again, they could choose among three options: never, sometimes, and 
often. As Table 35 illustrates, more respondents in the high treatment zones in Chad think 
that violence is an effective method to solve problems, while less respondents in the high 
treatment zones in Burkina do so. These results are different from those in Section B, 
which found that the P-DEV II program has no effect on this vulnerability indicator. There 
are no pooled effects in either low treatment or high treatment zones, as the positive 
effects in some countries and the negative ones in others cancel each other out.
Table 35: Violence Is Effective to Solve Problems 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.470 1.308 1.352 X
Baseline Low Core versus 
Non-Core .013 .088 .155 X 
Baseline High Core versus 
Non-Core -.190** .054 .219*** X 
Baseline-Midline 
Difference, Non-Core -.272*** .003 -.151*** X 
Difference in Differences, 
Low Core versus Non￾Core
-.040 -.162 .050 -.049 
Difference in Differences, 
High Core versus Non￾Core
.178* -.188 -.219** -.047 
* p < .10; ** p < .05; *** p < .01 
Next, we asked respondents whether they disagree or agree with the statement that 
“violence in the name of Islam” can be justified. As Table 36 illustrates, the notion that 
violence in the name of Islam can be justified has become more popular in Burkina’s high 
treatment zones relative to its no treatment zones, which suggests a detrimental impact 
of the P-DEV II program in this country. These results are different from those in Section 
B, which suggest that the P-DEV II program has no effect on this vulnerability indicator, 
and seems to run against the pattern seem thus far of positive program impacts on 
indicators of extremism in high intensity Burkina zones. Again, there is no pooled 
treatment effect in either of the zones. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
61 
Table 36: Violence in the Name of Islam 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.539 1.574 1.540 X
Baseline Low Core versus 
Non-Core .011 .052 -.085 X 
Baseline High Core versus 
Non-Core -.102 -.025 -.122 X 
Baseline-Midline 
Difference, Non-Core -.461*** -.395*** -.411*** X 
Difference in Differences, 
Low Core versus Non￾Core
.002 -.094 .125 -.001 
Difference in Differences, 
High Core versus Non￾Core
.115 .085 .173* .123 
* p < .10; ** p < .05; *** p < .01 
To measure anti-West attitudes, respondents were asked whether they disagree or agree 
with the following statement: “The United States is at war against Islam, not terrorism.” 
Table 37 shows that in both Chad and Niger, the notion that the U.S. is at war against 
Islam has become significantly less popular over time; this goes for low and high 
treatment zones alike. In Burkina, the opposite has been happening, with this notion 
having become more popular in both low and high treatment zones. These results are 
consistent to those in Section B, which suggest that the P-DEV II program decreases anti￾West attitudes in Chad in Niger but increases them in Burkina. It also suggests that there 
are few differences in these impacts across low and high intensity treatment zones. 
Table 37: Anti-West Attitudes 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.398 2.061 1.659 X
Baseline Low Core versus 
Non-Core .094 .245 -.095 X 
Baseline High Core 
versus Non-Core -.148*** .322** .125 X 
Baseline-Midline 
Difference, Non-Core .019 -.037 -.440*** X 
Difference in Differences, 
Low Core versus Non￾Core
-.293** -.636** .309*** -.221*
Difference in Differences, 
High Core versus Non￾Core
-.269*** -.986*** .146*** -.320 
* p < .10; ** p < .05; *** p < .01 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
62 
Goal 3: Youth Outlook 
Tables 38 through 42 show the results for questions related to Goal Level Index 3: Youth 
Outlook, defined as the individual and collective vision young people have of their futures. 
This concept is operationalized using three categories of indicators: a) economic outlook 
which evaluates attitudes toward and practical vision of future careers and economic 
potential; b) educational outlook which evaluates expectations regarding education and 
the learning environment; and c) civic outlook which comprises participation in civil society 
and local decision making. The tables in this section present the results for the full sample 
in each country, that is, individuals of all ages, given that few differences were found in 
the less statistically robust youth-only analyses in the previous section. 
Life Satisfaction and Economic Outlook (higher values represent improved outlook) 
To measure life satisfaction, respondents were presented with a card showing a ladder 
representing the “ladder of life.” The ladder consists of eleven steps, ranging from 0 to 
10. Respondents were then asked the following question: “Let’s suppose the top of the 
ladder is the best possible life for you; and the bottom, the worst possible life for you. On 
which step of the ladder do you personally stand at the present time?” As Table 38 
illustrates, there has been a significant and very large decrease in life satisfaction in 
Niger’s high treatment zones relative to its no treatment zones over time, with no changes 
attributable to the P-DEV II program elsewhere. 
Table 38: Life Satisfaction 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 4.624 3.842 4.250 X
Baseline Low Core versus 
Non-Core -.167 .606* -.060 X 
Baseline High Core versus 
Non-Core .300 .843*** .064 X 
Baseline-Midline 
Difference, Non-Core -.601*** .886*** .422 X 
Difference in Differences, 
Low Core versus Non￾Core
.064 -.330 .419 .068 
Difference in Differences, 
High Core versus Non￾Core
-.146 -1.068** .385 -.322 
* p < .10; ** p < .05; *** p < .01 
To measure perceptions of the economy, respondents were asked whether the economy 
of their country is worse, about the same, or better than it was a year ago. Table 39 shows 
that perceptions of the economy have significantly worsened over time in the high 
treatment zones relative to the non-core zones in Chad and Niger, a finding that is 
different from what was observed in Section B. Conversely, perceptions of the economy 
have significantly improved between waves in the high treatment zones in Burkina. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
63 
Table 39: Economic Outlook 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.080 2.191 2.373 X
Baseline Low Core versus 
Non-Core .027 .048 -.325** X 
Baseline High Core 
versus Non-Core .001 .110 -.669*** X 
Baseline-Midline 
Difference, Non-Core -.358** .011 -.293*** X 
Difference in Differences, 
Low Core versus Non￾Core
.057 -.076 .226 .049 
Difference in Differences, 
High Core versus Non￾Core
-.443*** -.292** .875*** .111 
* p < .10; ** p < .05; *** p < .01 
Educational Outlook (higher values represent improved outlook) 
We measure respondents’ attitudes regarding educational inclusiveness by asking 
whether they agree or disagree with the following statement: “Middle school is only for 
boys, not girls.” As Table 40 illustrates, attitudes regarding education have become more 
inclusive over time in Chad’s high treatment zones. These results are similar to those in 
Section B, which suggest that the P-DEV II program has a positive impact on educational 
inclusiveness in Chad. 
Table 40: Inclusiveness of Schools 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 2.644 2.601 2.615 X
Baseline Low Core versus 
Non-Core -.139 .020 .105 X 
Baseline High Core versus 
Non-Core -.015 -.038 .056 X 
Baseline-Midline 
Difference, Non-Core -.049 .184** .294*** X 
Difference in Differences, 
Low Core versus Non-Core .163 -.100 -.134 -.027 
Difference in Differences, 
High Core versus Non￾Core
.248** -.021 -.077 .056 
* p < .10; ** p < .05; *** p < .01 
Civic Outlook (higher values represent improved outlook) 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
64 
Political interest is by asking respondents whether they have very little interest, some 
interest, or a great deal of interest in local community affairs. Table 41 shows that interest 
in community affairs has significantly increased over time in the high treatment zones in 
Chad and Burkina. Conversely, interest in community affairs has significantly decreased 
between waves in the high treatment zones in Niger. These results are different from 
those in Section B, which suggest that the P-DEV II program has no impact on political 
interest. 
Table 41: Interest in Community Affairs 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core 1.577 1.651 2.196 X
Baseline Low Core versus 
Non-Core .060 -.088 -.270 X 
Baseline High Core 
versus Non-Core .086 -.142* -.350** X 
Baseline-Midline 
Difference, Non-Core .206*** .638*** .265* X 
Difference in Differences, 
Low Core versus Non￾Core
-.075 .106 .123 .023 
Difference in Differences, 
High Core versus Non￾Core
.330*** -.352*** .510*** .069 
* p < .10; ** p < .05; *** p < .01 
To measure this dimension, respondents were asked two questions: a) “Do you know 
how long the term of office is for the President?” b) “Do you know how many seats there 
are in the National Assembly?” The analyses in Table 42 are based on the average 
number of correct answers given across the two items. As the table illustrates, Niger and 
Burkina have experienced a significant increase in political knowledge over time in both 
low and high treatment zones. These results are similar to those in Section B, which 
suggest that the P-DEV II program has a positive impact on political knowledge in Niger 
and Burkina Faso. The overall pooled treatment effect is also significant for high intensity 
zones. 
Table 42: Political Knowledge 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Baseline, Non-Core .229 .479 .331 X
Baseline Low Core versus 
Non-Core .124*** -.066 -.008 X 
Baseline High Core versus 
Non-Core .004 .025 .097*** X 
Baseline-Midline 
Difference, Non-Core .094 -.032 .012 X 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
65 
Chad Niger Burkina 
Faso 
Pooled 
Treatment 
Effect
Difference in Differences, 
Low Core versus Non￾Core
-.033 .167** .086** .056 
Difference in Differences, 
High Core versus Non￾Core
.014 .189*** .059* .106***
* p < .10; ** p < .05; *** p < .01 
In sum, these analyses add nuance to the basic results shown in Section III-B above, but 
do not change the essential contours of the results already presented. As in the earlier 
section, there are few significant pooled cross-country effects: high intensity zones with 
greater amount of P-DEV II activities showed greater positive change on perceived 
access to jobs and political knowledge, relative to non-core zones, and low intensity 
zones showed greater declines in perception of the justifiability of religious violence and 
support for interethnic marriage. Examination of the country-by-country effects showed 
greater impact of the program in high intensity zones on a wider range of outcomes, 
especially in Burkina and especially on indicators of Resilience to Violent Extremism. This 
indicates that zones within Burkina with more P-DEV II activities showed the most 
consistent pattern of positive changes across the 19 indicators. Treatment intensity was 
generally less consequential for predicting positive program outcomes in Chad and Niger. 
Even in Burkina, however, there were anomalous findings where P-DEV II high intensity 
zones changed in ways opposite to the program’s goals. 
D. Panel Analysis 
A portion of the data collection exercise consisted of a panel study in which some 
individuals from EAS baseline (wave 1) communes in Niger and Burkina Faso were re￾interviewed in the midline survey (wave 2). The panel sample consists of 497 respondents 
who were randomly selected from the pool of wave 1 respondents, roughly 25 individuals 
per commune for 10 communes each in Niger and Burkina Faso. As noted above, 
logistical and budgetary issues prevented the panel portion of the study from being carried 
out in Chad. Surveying the same individuals allows us to examine the effects of the P￾DEV II program on the various outcome measures more directly. To do so, respondents 
in wave 1 and wave 2 were asked whether or not they participated in any (and if so, how 
many times) of the following seven activities during the past year: 
 A workshop or forum in your community where issues about peace or democracy 
were discussed 
 A poetry reading or a drama about peace or democracy issues 
 Training to develop ways to solve community problems 
 A workshop or forum at a place other than a mosque where imams or religious 
leaders discussed peace or democracy with people in the community 
 Trainings to help you find work or find a better job 
 Trainings to help you learn how to run a business or write an application for a 
business development grant 
 A meeting to help determine how community grant funding will be spent 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
66 
These activities were selected to represent the range of activities implemented by P-DEV 
II program. However, it is important to note that we do not ask specifically about exposure 
to P-DEV directly, since the sponsorship of the activities is rarely made known to 
participants. We cannot therefore say with certainty that positive responses to these 
questions indicate direct exposure to P-DEV II programming, since there may have been 
other civil society or donor-sponsored programs on these themes in the given zones. We 
use the change in these activities at the individual-level to control for individuals’ baseline 
level of exposure to other programs, and assume that at least some of the positive 
reported changes in exposure in core zones can be attributed to P-DEV II. 
First, to get a general sense of the extent to which respondents participated in these 
activities, we created a binary index coded “0” for individuals who did not participate in 
any of these seven activities during the past year, and “1” for individuals who participated 
in at least one. Figures 21 and 22 show the percentages of panel respondents coded as 
“0” and “1,” broken down by core/non-core zone and wave 1/2. As the graphs illustrate, 
there is a substantial difference between the two countries, with about 70% of wave 1 
respondents in Burkina having participated in at least one activity but only about 30% of 
wave 1 respondents in Niger having done so. The figures also show that in both countries, 
there are virtually no differences between core and non-core zones for the first wave. In 
Niger, the number of respondents who had participated in at least one activity had 
increased to about 40% in the non-core zones and some 48% in the core zones by the 
time of the midline survey. In Burkina, the number of respondents who had participated 
in at least one activity had decreased to 50% in the non-core zones and some 64% in the 
core zones by the time of the midline survey. Therefore, the increase in respondents who 
reported to have participated in at least one activity in Niger is greater for the core zones 
than the non-core zones, and the decrease in respondents who said they had participated 
in at least one activity in Burkina is smaller. As noted, we cannot be certain about the 
source of these reported changes in activity, but it seems likely that this overall increased 
participation in workshops, job trainings, etc. in the core zones relative to the non-core 
zones reflect at least to a significant degree the implementation of the P-DEV II program. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
67 
Figure 21: Binary Activity Indicator for Niger 
Figure 22: Binary Activity Indicator for Burkina Faso 
Additional evidence for this assertion comes from looking at how many times respondents 
participated in these activities, not just whether they did at all. To this end, we created an 
ordinal variable for each of the seven activities, coded “0” if the respondent had not 
participated in it during the past year, “1” if they did once or twice, “2” if they did three or 
four times, and “3” if they did more than four times. We then generated an additive index 
summing the ordinal variables for all seven activities. Next, we broke down this index by 
core vs. non-core and by wave 1 vs. wave 2. In non-core zones, the mean value for wave 
1 is 2.61, and the mean value for wave 2 is 1.62. In core zones, the mean value for wave 
1 is 1.62, and the mean value for wave 2 is 1.93. These numbers suggest that in P-DEV 
0 20 40 60 80 100 percent
Wave 1 Wave 2
Non-core Core Non-core Core
0 activities 1 or more activities
0 20 40 60 80 100 percent
Wave 1 Wave 2
Non-core Core Non-core Core
0 activities 1 or more activities
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
68 
II core, individuals participate more in the types of activities listed above. In sharp contrast, 
the non-core zones experienced a rather steep decline in participation in the time between 
wave 1 and wave 2. 
To shed further light on the extent to which individuals have participated in activities 
related to the P-DEV II program, we divide respondents into three categories: those who 
have participated in fewer activities in wave 2 relative to wave 1, those who have 
participated in the same amount, and those who have participated in more activities in 
wave 2 relative to wave 1. Figure 23 illustrates the percentage of respondents who fall 
into each category, broken down by country and zone type. It shows that in the core 
zones, the percentage of respondents who have participated in more activities is greater 
than it is in the non-core zones, and the percentage of respondents who have participated 
in fewer activities is smaller. This difference between core and non-core zones is 
especially pronounced for Burkina. 
Figure 23: Change in Program Participation 
In the next section, we turn our attention to the effects of participating in these seven 
activities on the 19 indicators used to operationalize the P-DEV II program goals. For 
each goal, we created a separate table, with each row representing one particular 
indicator. The coefficients represent the effect of the difference in program participation 
(i.e., the number of activities participated in during the year prior to the midline survey 
minus the number of activities participated in during the year prior to the baseline survey) 
on the respective indicator, controlling for the prior value of the outcome variable as well 
0 20 40 60 80 100 percent
Niger Burkina
Non-core Core Non-core Core
Less activities No change
More activities
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
69 
as baseline differences between Niger and Burkina.5
 This means that we attempted to 
link the changes in outcomes at the individual level from baseline to midline with the 
changes in exposure to the activities sponsored by P-DEV II (though without having 
absolute certainty that these changes are from the P-DEV II program). Again, for the 
reader’s convenience, the coefficients are color-coded: coefficients representing a 
desired effect of the program are shown in green, whereas coefficients representing a 
detrimental effect of the program are shown in red. The analyses were run two separate 
times: once for all zones, and once for core zones only to maximize the chances that the 
activities respondents participated in were actually part of the P-DEV program. 
Goal 1: Social Cohesion 
Table 43 shows the results for the panel analysis of the indicators relating to Goal 1: 
Social Cohesion. The table illustrates that the majority of coefficients for these indicators 
are small and size and statistically significant, suggesting that participating in a greater 
number of activities over time has no effect on them. There are two exceptions to this 
pattern: institutional trust in Burkina and political participation in Niger (only in the “all 
zones” specification). However, the former effect goes in the wrong direction: with each 
individual activity a respondent has participated in, their average level of trust in the five 
institutions mentioned above decreases by .011 (.028) points on a three-point scale. 
Conversely, for every additional activity a respondent has participated in, their average 
level of engagement in the three different types of political participation increases by .020 
points on a two-point scale. Overall, there is little difference between the results for the 
“all zones” specification and the “core only” specification. 
Table 43: Panel Analysis for Goal 1: Social Cohesion 
Niger 
(all zones)
Niger 
(core only)
Burkina 
(all zones)
Burkina 
(core only)
Interpersonal Trust .007 .009 .003 -.006
Institutional Trust -.006 .002 -.011* -.028**
Community Decision￾making .003 -.017 -.006 .021 
Political Participation .020*** .012 .001 .015
Interethnic Marriage -.003 .004 -.005 -.009
* p < .10; ** p < .05; *** p < .01 
Goal 2: Resilience to Violent Extremism 
Table 44 shows the results for the panel analysis of the indicators relating to Goal 2: 
Resilience to Violent Extremism. Many of the coefficients for these indicators are small in 
size and statistically insignificant. This is especially true for Burkina, where increased 
participation in activities over time affects only one (two) of the indicators. For Niger, 
several coefficients relating to the resilience and vulnerability indicators are significantly 
influenced by the difference in program participation, but most of them suggest a 
detrimental effect of the P-DEV II program. Again, the results for the panel analyses 
remain largely the same, regardless of whether they are run on all zones or on the core 
 
5
 The complete regression tables are included in the Appendix. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
70 
zones only. 
Table 44: Panel Analysis for Goal 2: Resilience to Violent Extremism 
Niger 
(all zones)
Niger 
(core only)
Burkina 
(all zones)
Burkina 
(core only)
Access to Jobs -.001 -.035* .005 .029*
Access to Vocational 
School .004 .019 .034** .026 
Political Efficacy .006 -.030 -.002 .004
Ethnic Differences .038** .044** .005 -.001
Religious Differences .041* .048** .007 -.001
Religious Violence 
Justified -.003 .020 -.001 -.001 
Violence Effective -.027* .018 .007 .002
Violence in the Name of 
Islam .011* .014 .007 -.001 
Anti-West Attitudes .022** .031* .014 .047**
* p < .10; ** p < .05; *** p < .01 
Goal 3: Youth Outlook 
Table 45 shows the results for the panel analysis of the indicators relating to Goal 3: 
Youth Outlook. Again, the table illustrates that the majority of coefficients for these 
indicators are small and size and statistically significant, suggesting that participating in 
a greater number of activities over time has no effect on them. Increased participation in 
activities over time appears to negatively affect respondents’ economic expectations in 
Burkina and attitudes toward educational inclusiveness in Niger. Conversely, it seems to 
have a positive impact on respondents’ life satisfaction in Burkina (only in the “core only” 
specification) and political knowledge in Niger (only in the “all zones” specification). Again, 
the coefficients for the “all zones” specification and the “core only” specification are largely 
the same in terms of direction and significance level. 
Table 45: Panel Analysis for Goal 3: Youth Outlook 
Niger 
(all zones)
Niger 
(core only)
Burkina 
(all zones)
Burkina 
(core only)
Life Satisfaction .004 -.035 .044 .198**
Economic Outlook -.002 -.024 -.025** -.044*
Inclusiveness of Schools -.014* -.031** .001 .009
Interest in Community 
Affairs -.005 .020 .007 -.002 
Political Knowledge .009* .004 .004 .002
*p < .10; ** p < .05; *** p < .01 
Unfortunately, the panel portion of the study was unsuccessful in linking program 
exposure to outcomes at the individual level. In almost every instance, individuals who 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
71 
reported greater exposure to the kinds of activities that P-DEV II implemented between 
baseline and midline showed either no difference over time on outcomes relevant to 
program goals or showed changes in the “wrong” direction. To this extent, we could not 
replicate the results of the zone-level analyses shown in earlier sections of the report at 
the individual level. This is likely due to the relatively few numbers of people who were re￾interviewed in each zone from the EAS baseline zones, which already were a limited 
number of zones in which the P-DEV II program was active. 
E. Experimental Evidence 
The indicators discussed in the previous sections share the common feature that they all 
request direct answers or “self-reports” concerning personal perceptions of violence that 
may occur in their communities, the extremist violence that may occur beyond their 
communes, and various justifications for using violence to solve problems or in the name 
of religion. However, researchers cannot often be confident that these types of questions 
yield honest or externally valid responses from respondents. 
The first concern about self-reported responses relates to the honesty of the response. 
In fact, it is well-known in public opinion and social psychology research that survey 
respondents tend to answer questions in a way that they believe will please enumerators 
or, more relevant to this report, provide desirable responses to sensitive items because 
they conform to a particular set of societal norms. The second concern is the external 
validity of the response. Are these findings generalizable to political problems in the real 
world? Because the self-reported items are framed in abstract terms, it is not clear they 
capture genuine intentions to use violence. That is, such items identify approval of violent 
behaviors for vaguely-defined circumstances, but they do not measure willingness to 
support specific violent reactions to clearly-defined policies. 
For these reasons, the EAS team devised innovative evaluation instruments that measure 
attitudes toward violent extremism unobtrusively, that is, minimizing the potential for 
untruthful answers. These techniques attempt to mitigate social desirability bias and 
improve internal validity of measures of vulnerability and resilience to violent extremism. 
They also improve external validity of measures by providing specific and realistic stimuli 
and relevant violent reactions. By doing so, these items capture the willingness to use 
violence in concrete, policy-based terms. 
List Experiment # 1 
The list experiment procedure is straightforward. Respondents are presented with a 
hypothetical scenario and a series of instructions, read to them by the survey enumerator: 
“Imagine that you hear on the radio that a Western newspaper has published offensive 
images of the Prophet Mohammed committing a crime. You could respond to this affront 
in many ways. I’m going to read you a list of possible responses now. Please listen to 
them and then tell me how many of the following reactions you would support.” 
The sentence in italics represents the core of the list experiment because respondents 
are instructed to reveal only the number of responses they would support but not which 
responses they would support. Half of respondents in a given target zone were randomly 
assigned to a “control group” to whom the survey enumerator lists only three potential 
responses: 
1) A peaceful protest at the Western country’s embassy; 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
72 
2) Your government demands an apology from the Western country; 
3) Your government declaring war against the Western country. 
Responses 1-2 are likely unobjectionable. The third item is designed to be more radical 
so that that most respondents do not necessarily respond affirmatively to all three control 
items. This is referred in the literature as a ceiling item or a low-prevalence item that 
minimizes design effects when comparing control and treatment groups. The other half 
of respondents in a target zone were randomly assigned to a “treatment group,” to which 
the survey enumerator lists the three potential responses listed above as well as a fourth, 
sensitive item: 
4) An attack on the Western country’s embassy that could result in military or civilian 
casualties. 
Again, respondents only reveal the total number of potential responses to the hypothetical 
scenario that they would support and they do not reveal to the survey enumerator which 
ones. Assuming that the randomization process for assigning treatment and control 
groups was effective, the difference between the average number of items that 
respondents in the treatment group report and the average number of items that 
respondents in the control group report therefore yields a measure of the percentage of 
the sample that agrees that “An attack on the Western country’s embassy that could result 
in military or civilian casualties” is an appropriate response to the publication of an 
offensive image of the Prophet Mohammed. 
For our analyses of the first list experiment, we model the respondent’s answer as a linear 
function of his/her treatment assignment and the control covariates. These OLS 
regressions predict the count of violent acts provided by respondents with a dichotomous 
variable that indicates whether or not the respondent received the sensitive item 
(attacking the embassy). Under the conditions of randomization, the treatment groups 
and control groups are equally likely to support all three non-sensitive items. Given that 
the groups are identical on every measure except for the addition of a fourth item on the 
list, any differences between the groups may be attributed to the addition of the sensitive 
item. Specifically, differences between the groups may be interpreted as the proportion 
of respondent who support an attack on an embassy that may result in military or civilian 
casualties. This difference between the treatment and control groups is referred to as the 
“average treatment effect.” 
Table 46: List Experiment #1 – Percentage of Respondents Who Choose the Sensitive Item 
Pooled Chad Niger Burkina Faso
Midline, Non-core 16.6*** 16.8** 19.3*** 13.6**
Midline , Core 18.6 17.4 23.5 12.3
* p < .10; ** p < .05; *** p < .01 
First, we analyze the responses by all individuals interviewed during the second wave of 
the survey. The results are shown in Table 46, broken down by country and zone type. 
The first row shows the proportion of respondents in non-core zones who support an 
attack on an embassy that may result in military or civilian casualties, and whether this 
number is significantly different from zero. In all three countries, a significant portion of 
respondents in non-core zones in all three countries supports such an attack. The table 
also reveals differences between the country, with Niger exhibiting the greatest portion of 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
73 
respondents who choose the sensitive item and Burkina exhibiting the least. The second 
row shows the proportion of respondents in core zones who support an attack on an 
embassy that may result in military or civilian casualties, and whether this number is 
significantly different from the results for the non-core zones. In all three countries, the 
difference is insignificant, meaning that midline respondents in core and non-core zones 
are statistically indistinguishable in terms of their level of support for such an attack. 
Second, we analyze and compare the changes in baseline and midline responses among 
individuals only in EAS zones, following the difference in difference logic utilized to 
estimate P-DEV II program effects in earlier portions of the report. The reason we restrict 
this analysis to respondents who are residing in a zone that was part of the baseline EAS 
survey is that during the baseline wave, respondents in zones surveyed by IRD were not 
asked the questions relating to the experiments. Therefore, it was not possible to measure 
changes among IRD zones from baseline to midline, nor to link these changes to core 
versus non-core P-DEV II status. 
Table 47: List Experiment #1 – Percentage of Respondents Who Choose the Sensitive Item 
Chad Niger Burkina Faso
Baseline, Non-core 10.9* 3.2 1.8
Baseline, Core 10.9 9.3 10.2
Midline, Non-core 6.0 17.5 8.6
Midline, Core 28.4 13.0 8.9
* p < .10; ** p < .05; *** p < .01 
Table 47 illustrates the results of the analysis of EAS zones over time. The first row shows 
the proportion of baseline respondents in non-core zones who support an attack on an 
embassy that may result in military or civilian casualties, and whether this number is 
significantly different from zero. Only in Chad, the proportion of respondents who support 
such an attack is large enough to reach statistical significance. In Niger and Burkina, only 
a small percentage of respondents choose the sensitive item; these numbers are so small 
that they are statistically indistinguishable from zero. The second row shows the 
proportion of respondents in core zones who support an attack on an embassy that may 
result in military or civilian casualties, and whether this number is significantly different 
from the results for the baseline non-core zones. In Chad, the figures for core and non￾core zones are virtually the same. In Niger and Burkina, the proportion of respondents 
who choose the sensitive item is larger in the core zones than it is in the non-core zones, 
but not large enough to reach conventional levels of statistical significance. 
The third row shows the proportion of midline respondents in non-core zones who support 
an attack on an embassy that may result in military or civilian casualties, and whether this 
number is significantly different from the results for the baseline non-core zones. In non￾core zones in Niger and Burkina, the proportion of respondents who support such an 
attack has increased over time, while Chad exhibits the opposite trend. This suggests 
increased support for violent acts against Western civilians in the former two countries 
and decreased support in the latter, although it should be added that these differences 
are statistically insignificant. The last row shows the proportion of midline respondents in 
core zones who support an attack on an embassy that may result in military or civilian 
casualties. In core zones Chad and Niger, support for such an attack has increased over 
time, while Burkina exhibits the opposite trend. However, even in Chad where the 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
74 
percentage of respondents who choose the sensitive item increases from 10.9 to 28.4, 
the difference does not reach conventional levels of statistical significance. 
List Experiment #2 
For the second list experiment, respondents are presented with another hypothetical 
scenario and a series of instructions, read to them by the survey enumerator: 
“Now to change the subject, I am going to read you a list of things that people in this area 
may have considered doing. Please listen to them and then tell me how many you would 
consider doing. Please, do not tell me which ones, only how many.” 
The sentence in italics represents the core of the list experiment because respondents 
are instructed to reveal only the number of responses they would support but not which 
responses they would support. Half of respondents in a given target zone were randomly 
assigned to a “control group” to whom the survey enumerator lists only four potential 
responses: 
1) Donating clothes or money to someone in need; 
2) Voting for a candidate who is a strong supporter of homosexual rights; 
3) Reporting an official who has taken bribes; 
4) Going on the Hajj. 
The other half of respondents in a target zone were randomly assigned to a “treatment 
group,” to which the survey enumerator lists the four potential responses listed above as 
well as a fifth, sensitive item: 
5) Joining a group that carries out acts of violence to defend your religion. 
Again, respondents only reveal the total number of potential responses to the hypothetical 
scenario that they would support and they do not reveal to the survey enumerator which 
ones. Assuming that the randomization process for assigning treatment and control 
groups was effective, the difference between the average number of items that 
respondents in the treatment group report and the average number of items that 
respondents in the control group report therefore yields a measure of the percentage of 
the sample that agrees that “Joining a group that carries out acts of violence to defend 
your religion” something that people in this area may have considered doing. 
To analyze the second list experiment, we again model the respondent’s answer as a 
linear function of his/her treatment assignment and the control covariates. The questions 
relating to the second list experiment were not included in the questionnaire during the 
baseline wave of the survey, which is why we cannot make any comparisons over time. 
Still, the second list experiment can provide us with important information about the extent 
to which terror groups may find potential recruits in each of the three country at the time 
of the midline survey. 
Table 48 illustrates the results of our analysis, broken down by country and zone type. 
The first row shows the estimated proportion of respondents in non-core zones who would 
consider joining a group that carries out acts of violence to defend your religion, and 
whether this number is significantly different from zero. In non-core zones in Chad and 
Burkina, virtually no respondents think that this is the case; in stark contrast, 41.5 percent 
of the respondents in non-core zones in Niger choose the sensitive item. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
75 
Table 48: List Experiment #2 – Percentage of Respondents Who Choose the Sensitive Item 
Pooled Chad Niger Burkina Faso
Midline, non-core 14.0* 0.0 41.5*** 0.0
Midline, core 11.7 8.2 6.8* 20.5
* p < .10; ** p < .05; *** p < .01 
The second row shows the proportion of respondents in core zones who may have 
considered joining a group that carries out acts of violence to defend your religion, and 
whether this number is significantly different from the results for the non-core zones. In 
Chad and Burkina, the percentage of respondents who do think that this is the case is 
higher in the core zones than in the non-core zones, but these differences are 
insignificant. In Niger, the percentage of respondents choosing the sensitive item is 6.8, 
which is substantially and significantly smaller than the 41.5 percent exhibited in the non￾core zones. 
Overall, the results of our second list experiment suggest that there is a non-negligible 
pool of individuals who may consider joining groups that would carry out acts of religious 
violence; this pool is largest in non-core zones in Niger and core zones in Burkina Faso. 
Endorsement Experiment 
We also implement an endorsement experiment that measures support for the extremist 
group Al Qaeda in the Islamic Maghreb. The endorsement experiment follows a similar 
structural logic as the list experiment, though its aim is to measure sympathy with active 
radical and extremist groups in the region, as opposed to approving of extremist violence 
(Bullock et al. 2011). 
For our endorsement experiment, respondents are randomly assigned to receive one of 
two prompts. In the control condition, the survey enumerator reads: 
“The World Health Organization recently announced a plan to introduce universal polio 
vaccinations across {Burkina Faso/Chad/Niger}. How much do you approve of such a 
plan – not at all, somewhat, or quite?” 
Treatment subjects, by contrast, encounter one additional detail: 
“The World Health Organization recently announced a plan to introduce universal polio 
vaccinations across {Burkina Faso/Chad/Niger}. It is likely that Al-Qaeda in the Islamic 
Maghreb (AQIM), an Islamist group, will oppose this program. How much do you 
approve of such a plan – not at all, somewhat, or quite?” 
The baseline expectation is that individuals in the treatment group who sympathize to 
some extent with AQIM will be more likely than individuals in the control group to oppose 
the WHO program. 
Table 49 displays the percentage of all midline respondents who report to disapprove of 
the nation-wide vaccination program. The control group contains the individuals who 
evaluate the vaccination program on its own, with no group endorsement heuristic. Of 
these individuals, opposition to the program ranges from 14.9% in non-core zones in 
Burkina to 38.2% in core zones in Chad. The treatment group contains individuals who 
evaluate the vaccination program with the additional knowledge that AQIM opposed the 
program. Of these individuals, opposition to the program ranges from 27.3% in non-core 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
76 
zones in Burkina to 62.2% in core zones in Chad. From these figures, two patterns 
emerge. First, there are significant differences between countries in terms of how 
receptive individuals are to cues from extremist groups, with respondents in Chad being 
the most receptive and those in Burkina the least. Second, such differences also exist 
between non-core and core zones, with respondents in the latter being more receptive to 
cues from extremist groups. 
Table 49: Endorsement Experiment #1 – Percentage of Midline Respondents Who Disapprove of 
Vaccination Program 
Chad Niger Burkina Faso
Non-core Core Non-core Core Non-core Core
Control Group 24.9 38.2 24.1 30.9 14.9 10.24
Treatment Group 54.1 62.2 49.8 56.3 27.3 36.31
* p < .10; ** p < .05; *** p < .01 
We also analyze the responses by individuals interviewed during the second wave of the 
survey who are residing in a zone that was part of the wave 1 EAS survey, and compare 
their responses to those by individuals interviewed during the first wave of the survey. 
Again, the reason we restrict this analysis to respondents who are residing in a zone that 
was part of the wave 1 EAS survey is that during the first wave of the survey, respondents 
in zones surveyed by IRD were not asked the questions relating to the experiments. The 
results are shown in Table 50. 
The first row shows how more likely baseline respondents in non-core zones are to 
oppose the program if it is also opposed by AQIM than if it is not, and whether this number 
is significantly different from zero. Respondents are roughly 11% (Chad), 22% (Niger), 
and 22% (Burkina Faso) percentage points more likely to oppose the program if it is also 
opposed by AQIM than if it is not. Following the logic of the experiment, these individuals’ 
decisions appear to reflect positive evaluations of the extremist group. 
Table 50: Endorsement Experiment #1 – DiD Analysis 
Chad Niger Burkina Faso
Baseline, non-core 11.4** 21.5*** 22.0***
Baseline, core 16.6 23.7 21.9
Midline, non-core 20.5 34.2* 12.6**
Midline, core 24.8 32.9 27.8
* p < .10; ** p < .05; *** p < .01 
The second row shows how more likely baseline respondents in core zones are to oppose 
the program if it is also opposed by AQIM than if it is not, and whether this number is 
significantly different from the results for the baseline non-core zones. In all three 
countries, the difference between non-core and core zones is insignificant, meaning that 
the differences between the two are negligible. The third row shows how more likely 
midline respondents in non-core zones are to oppose the program if it is also opposed by 
AQIM than if it is not, and whether this number is significantly different from the results 
for the baseline non-core zones. In non-core zones in Niger, the percentage of 
respondents who base their decision on a positive evaluation of the extremist group has 
increased significantly 34.2%, while in Burkina this number has decreased to 12.6%. The 
fourth row shows how more likely midline respondents in core zones are to oppose the 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
77 
program if it is also opposed by AQIM than if it is not, and whether the DiD is statistically 
significant. In all three countries, positive evaluations of the extremist group have 
increased over time in the core zones, but these differences do not reach conventional 
levels of statistical significance. Therefore, while we can say that there was a general 
increase in support for AQIM in Niger and a general decrease in Burkina, we cannot 
attribute these changes to the activities carried out by the P-DEV II program. 
IV. Observations from Qualitative Data 
This section outlines the common themes and key commentary from the qualitative data 
collected by CERFODES in Burkina Faso and CASPA in Niger. The overall context 
surrounding P-DEV II activities in the two countries presents deteriorating conditions 
within the two countries. However, key differences exist, which support the country-by￾country results in the DiD analysis above. Burkina Faso showed positive attitudes toward 
social cohesion and resilience to violent extremism, particularly in perceived ethnic and 
religious differences and rejection of violence from extremists. Niger showed negative 
attitudes toward social cohesion, particularly in religious tensions between Islamic sects. 
In resilience to violent extremism, some respondents accepted activities of Al Qaeda but 
not Boko Haram, showing support for some forms of violent extremism. Other notable 
differences stand out across the two countries, particularly in terms of women’s autonomy 
and the effectiveness of P-DEV II activities. 
Both Burkina Faso and Niger cited weakening conditions that support youth outlook: 
employment, attraction to extremism and conflict, and ease of manipulation. In general, 
many of the themes that emerged during this phase of qualitative interviews and focus 
groups reinforce findings from the quantitative data. 
The methodology for qualitative data collection centered on focus group discussions with 
groups of seven to twelve participants. Groups were intentionally varied in composition, 
in order to include the views of a range of local residents: youth, educated women, Muslim 
men, young women, rural men, illiterate rural women, etc. Importantly, the research teams 
employed a strategy of in vivo coding, whereby facilitators started with questions related 
to the core themes of the research but then allowed sub-themes to emerge naturally 
through participant responses. Interviews with individuals complemented the focus 
groups; in each country, the partner organizations conducted ten focus groups and fifteen 
interviews. 
A. Burkina Faso 
Commentary was largely optimistic. Several of the focus groups finished by noting that 
they are hopeful in the future of their communities and that they foresee the current 
challenges facing their communities eventually being resolved. 
Exposure to P-DEV II 
P-DEV II programs were not well known to all groups, but the groups that were familiar 
with the programs thought highly of them, noting contributions to: peace; sensitization; 
positive behaviors; local governance; conflict resolution; consultations; discussion of 
problems in the community; and resource management. A trend in the data is that focus 
groups cited the importance of trainings for young people. Media messaging was cited 
often as well: Participants in Ouagadougou, Ouahigouya, and Gorgajdi cited participatory 
theater and cinema, while in Dori, Gorgajdi, and Ouahigouya, participants mentioned the 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
78 
role of radio programming. 
Respondents found that P-DEV II Community Action Committees in Ouagadougou, 
Ouahigoya, and Seguenega and theater groups in Seytenga contributed to conflict 
resolution. In Ouagadougou, a group of young males shared the following sentiment: 
We know the P-DEV II programs. They contribute to peace here, which is a tremendous 
contribution. The program especially supports associations that raise awareness. 
However, there is less familiarity with the programs in other areas, and sensitivity around 
the programs exists. A respondent expressed a concern about the identity of P-DEV II, 
which is seen as “American” and which may thus handicap the program. A respondent in 
Dori expressed a concern that there is no incentive associated with the program for 
residents to actually take action; they are focused instead on awareness. A respondent 
in Gorgajdi expressed concerns about the weakness of the radio grid, which hampered 
the effectiveness of the radio programs. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
79 
Goal 1: Social Cohesion 
Systematic communal tensions and conflict do not arise as a function of religious 
differences in Burkina. In many zones, respondents were also quick to point out that 
ethnic conflict does not frequently exist, as noted in Markoye, Seguenega, Tougo, and 
Yatenga. 
Instead, to the extent that communal tensions persist, the roots seem to be in competition 
between agriculturalists and pastoralists. Members of the Fulani (Peuhl) ethnicity are the 
principally pastoralists, so they are often involved. Because of this overlap in livelihood 
activities and ethnicity, however, those tensions often take on an ethnic guise. 
Furthermore, because the Fulani people are typically Muslim, the same competition over 
land can aggravate religious tensions. 
Resource conflict and disputes between pastoralists and sedentary farmers occur 
especially during periods of grazing shortage, as noted by respondents in Ouagadougou, 
Markoye, Tougo, and Yatenga. Violence associated with mining and group conflict over 
jobs was also noted by participants in Ouahigouya, Gorgajdi, Markoye, and Yatenga. 
Women and men clearly do not see the contributions of women in the same way: women 
cite their “enormous responsibilities,” whereas men see the role of women in a more 
simple and straightforward way. Male respondents in Burkina Faso do articulate a fairly 
strong recognition of the importance of female autonomy, however. Respondents in 
Ouagadougou noted that domestic violence occurs between men and women over the 
retrieval of water. 
Tellingly, to the extent that women do enjoy empowerment and autonomy, those efforts 
seem restricted to taking care of the family. Several male respondents in Namissiguima 
expressed the following: 
Women now hold jobs to support their families. In reality women have incomes but often 
it is to look after their families as their husbands have abandoned that to improve their 
personal living conditions. It is difficult to talk about women's empowerment in these 
conditions. 
Respondents expressed widespread frustration with and lack of trust in political leaders 
and local government. Conversely, several focus groups expressed that they “put faith in 
their traditional and religious leaders to help resolve local problems.” This is particularly 
true regarding conflict management resources and strategies for addressing land 
disputes, as expressed by focus group participants in Ouagadougou, Ouahigoya, and 
Seguenega, and by local officials interviewed in Seguenega, Seytenga, and Yatenga. 
Respondents also cited the use of religious authorities to help build social cohesion in 
Seguenega, and noted that women made a contribution to cohesion in Ouahigouya. 
Despite the absence of overt religious conflict in Burkina, inter-religious intolerance has 
posed a challenge to social cohesion. Respondents in Ouagadougou noted that churches 
are often noisy and disruptive in places where churches and mosques are located near 
each other. Muslims can cause difficulty for Christian groups too, as noted in Yatenga, 
where Muslims of certain sects do not accept the offer of Christians’ food during holidays. 
This, in turn leads the Christians to reject the Muslims. 
Goal 2: Resilience to Violent Extremism 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
80 
Groups expressed a strong sentiment against extremism. Terrorist groups are 
systematically viewed as incompatible with Islam, and respondents see no meaningful 
difference between Boko Haram and Al Qaeda. Respondents in Markoye, Ouagadougou, 
Ouahigouya, Seytenga, Tougo, and Yatenga, thought of these groups as equally evil. 
Respondents in Seytenga noted that they reject all forms of terrorism. Muslim 
respondents in Namissiguima unanimously expressed the following: 
Islam contributes to the promotion of peace…There is no difference between Al Qaeda 
and Boko Haram. They are not Muslims because the things they do (killing, assassination, 
rape, robbery) are condemned by Islam. 
There is growing concern over influence from across borders in terms of extremist threats. 
Participants in Ouagadougou, Ouahigouya, Seytenga, and Markoye were concerned 
about threats from outside of the country. Respondents from Ouahigouya, Seytenga, 
Markoye, and Yatenga cited terrorist influences from Niger, Mali, and across the Sahel, 
noting that being geographically close to extremists in other countries threatened their 
safety. 
Furthermore, the women’s group in Ouagadougou expressed growing fears of religious 
violence because “we don’t know who is who” (in reference to the ability of extremists to 
blend into the community). Respondents in Seytenga refused to integrate an Islamic sect 
from Niger out of concerns of terrorism. However, not all communities felt threatened by 
exposure to terrorist threats. Respondents in Gorgajdi, Markoye, Ouahigouya, and Tougo 
found security to be acceptable in their communities. 
Goal 3: Youth Outlook 
In Burkina Faso, respondents noted that youth were negatively affected by a lack of 
political integration, dominance by elders, and lack of authority. Respondents in 
Ouagadougou cited a lack of authority, monitoring, and mentoring, and in both 
Ouagadougou and Seytenga, the neglect of youth evoked animated discussion. There 
was disagreement over whether youth are neglected among respondents in these zones 
and those in Ouahigouya. 
A consistent concern is that the youth of Burkina Faso are “easy to manipulate.” This is 
especially true of youth who lack opportunities. Some suggest a rising tide of delinquency 
among young people, and that youth who lack steady activities, jobs, or opportunities are 
more likely to fall victim to the temptations of joining in illicit activities. 
The drivers that affected attraction to extremism and conflict among youth were similar 
across focus groups and interviews. Respondents perceived that youth were attracted to 
extremism by easy access to money in Ouagadougou and Yatenga, and by the lack of 
economic alternatives in Ouagadougou. Concerns relating to unemployment included 
idleness, as noted in Ouahigouya, Markoye, and Seytenga, and unease and nervousness 
among youth who do not work, cited by respondents in Markoye. 
Respondents expressed related concerns about the drivers of youth conflict, including 
vulnerability and lack of jobs or employment, as noted by participants in Ouahigouya and 
Yatenga, and attraction to easy money, noted in Yatenga. 
Respondent perceptions of youth attraction to extremism extended beyond economic 
drivers to neglect, weak faith, lack of authority or oversight, ease of manipulation and bad 
governance. Respondents in Ouagadougou cited weak religious faith, and in 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
81 
Ouahigouya, respondents widely agreed that youth were persuaded by Muslim preachers 
“who brainwash” youth. 
Governance also influenced youth attraction to extremism, as noted by respondents in 
Ouagadougou. Shortcomings in education and illiteracy, they argued, can lead to the 
ease of indoctrination. 
B. Niger 
Commentary was notably less optimistic in Niger. Respondents devoted more attention 
to problems and were more likely to assign blame to targeted actors. In particular, 
responses in Niger were generally less favorable to the West and American interests, 
particularly concerning terrorism related questions. 
Exposure to P-DEV II 
P-DEV II programs were known to several of the groups. Respondents who were familiar 
with P-DEV II interventions were appreciative of the program, citing the contributions of 
vocational training to small- and medium-sized enterprises and to economic activity; they 
also highlighted the awareness campaigns as a positive outcome of the program. 
Respondents in Balleyara, Diffa, Tahoua, and Zinder noted vocational training. 
Respondents in Tillaberi acknowledged support for Koranic Schools and the distribution 
of resources to schools, as well as participatory theater and mobile cinema. In Zinder, 
respondents stressed the value of workshops and trainings, and in Tahoua respondents 
cited radio broadcasts, lectures, and sports and cultural activities. 
However, a consistent problem cited by respondents in Niger was that the programs do 
not reach enough people to really be effective. Respondents noted limitations on reaching 
youth, where “a minority of beneficiaries” was perceived as having access to the program. 
One respondent expressly criticized the program, claiming that little consultation was 
occurring. Respondents in Zinder expressed the following: 
The P-DEV II workshops and trainings are effective programs in fighting against 
extremism and violence. The problem is that the programs do not reach a large enough 
number of people; they need to be expanded. 
Goal 1: Social Cohesion 
Respondents from several areas pointed out that ethnic conflict does not exist in their 
communities, including in Agadez, Balleyara, Diffa, Maradi, Tillaberi, and Zinder. However 
in Arlit, respondents described regular conflict between the Hausa and Tuareg ethnicities. 
Respondents noted that direct clashes rarely occur. 
Men expressed a more conservative view of women’s autonomy in Niger compared to 
Burkina Faso. One group of men, in Maradi, expressed that it is the man’s role to set 
limits on the woman, while the women’s group in Balleyara stated that women “are treated 
as inferior.” Groups regularly cited that the women’s place is in the home, and there 
seemed to be little enthusiasm for women engaging in public matters. 
Respondents identified a number of local conflict management resources: civil society, 
noted in Agadez, Niamey, and Tahoua; local authorities, noted in Agadez, Balleyara, and 
Zinder; elected officials, noted in Maradi and Zinder; and religious leaders, noted in 
Niamey, Tahoua, and Zinder. A large number of respondents stressed the role of women 
as potential mediators; this view was expressed in Agadez, Ballyara, Diffa, Maradi, 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
82 
Tahoua, and Tillaberi. 
Respondents in Niger asserted that Islam is a religion of peace that helps to promote 
social awareness. They were more favorable toward Sharia law and viewed Koranic 
schools and imams more favorably than did respondents in Burkina. The group of youth 
in Arlit noted that Sharia and active imams can help to “reduce the influence of the West,” 
which they viewed as a positive trend. Another group, in Tillabery, noted that Sharia can 
serve as a way to combat favoritism and injustice. Sharia was a solution that participants 
perceived to be favored by youth to combat injustice, noted in Arlit, Tillaberi, and Zinder. 
Differences within Islam weaken social cohesion in Niger. Those differences include 
opposing interpretations of the Koran by sects with distinct views—between the Izala and 
Tidjania in Balleyara, for example, and between Darika followers and Izala Sunnites in 
Zinder. Respondents in Tillaberi acknowledge that increasing religious fundamentalism is 
a factor in the expanding threat of Islamic terrorism. However, a Christian preacher in 
Tillaberi noted that his group is welcomed by Muslims, and he preaches coexistence. 
There is some sentiment among the respondents in Niger that the United States is in a 
war against Islam. A focus group of men in Diffa cited as evidence the fact that the U.S. 
only seems to attack countries that are majority Muslim. The Tillabery group expressed a 
similar opinion. Challenges to Western influence extend beyond the United States’ battle 
with terrorism: respondents in Maradi also noted that sometimes religious leaders boycott 
practical, Western-led initiatives such as vaccination campaigns. 
Goal 2: Resilience to Violent Extremism 
Respondents acknowledge that adaptation to terrorism has become “a compulsory 
reality” in the country. Even respondents located in areas that, so far, have not been 
affected by extremist violence, such as Tahoua, note that terrorism is “a true threat even 
though it hasn’t occurred here.” 
In Niger, focus groups cited a difference between Boko Haram and Al Qaeda. For 
example, in Tahoua and Tillaberi, Al Qaeda was described as acting for “political reasons” 
or, according to one group, “combatting miscreants.” Conversely, Boko Haram was 
described as “barbarous,” particularly because they kill Muslims. It should be noted that 
local Nigeriens have frequently been targeted by Boko Haram in the time period since the 
baseline data collection in 2013. Residents of Burkina Faso have not been targeted in the 
same way; the terrorist activity there has largely been aimed at foreign (Western) targets. 
This likely explains the difference of opinions across the two countries regarding 
comparisons of Al Qaeda and Boko Haram. 
Respondents cite cross-border extremist influences as a real problem, as in Burkina 
Faso. A respondent in Niamey noted that terrorism has “relocated to Niger.” This was 
particularly true in Maradi and Diffa, which is close to the border with northeastern Nigeria 
and a frequent target of Boko Haram attacks. Respondents there also expressed unease 
regarding refugees from Nigeria who lack identity papers, suggesting an increased 
vulnerability from those coming across the border. Across Niger, concerns of regional 
terrorist threats were palpable; a respondent in Arlit noted that the geographic proximity 
of Niger to Mali, Algeria, and Libya posed a particular threat to safety. 
Goal 3: Youth Outlook 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
83 
Regarding youth, respondents roundly expressed that youth are disorganized, 
dependent, and neglected, though opinions differed as to why. Some suggested that 
youth are subject to discrimination. Respondents in Arlit, Balleyara, Maradi, and Niamey 
noted that youth lack autonomy and face abandonment. This isolation extends to 
consultation on decision-making and lack of political opportunities, acknowledged by 
respondents in Maradi. 
Economic challenges to youth included the perception among respondents that youth 
lack opportunities and do not effectively take advantage of the few opportunities they do 
have, as noted in Balleyara, Diffa, Maradi, and Tillaberi. Youth widely face unemployment 
and joblessness, noted by respondents in Agadez, Balleyara, Tahoua, Tillaberi, and 
Zinder, as well as poverty, noted in Diffa, Maradi, and Zinder. 
Respondents found that youth participation in extremism and conflict are linked to 
unemployment and drug consumption. They also offered details regarding the links 
between youth and conflict, including: disagreements over women, poverty, and the need 
for belonging, as noted in Tillaberi, Zinder, Diffa, and Maradi. Youth become involved in 
political disputes when soliciting assistance from wealthy patrons, a problem noted 
Niamey. This leads to participation in the political opposition, cited as a source of youth 
participation in conflict. 
Respondents also expressed concerns about drugs and also about unwarranted 
harassment from police, particularly in Agadez. Furthermore, several groups noted that 
the youth “have a poor grasp of Islam” and lamented the declining influence of Imams 
among young people. Female respondents in Balleyara tied together some challenges 
facing youth in the following way: Young people here lack opportunities and end up just 
loitering around with nothing to do. They are also sometimes favorable to extremism 
because of their ignorance regarding Islam. 
C. Summary 
The results from the qualitative data lend insight into worsening conditions in the two 
countries. The context surrounding the P-DEV II program goals has shifted since the 
collection of baseline data in 2013. These findings support the DiD analysis, as youth 
outlook worsened and increased violent extremist activity affected attitudes of support, 
particularly in Niger. Positive attitudes toward social cohesion and resilience to violent 
extremism are found in Burkina Faso while in Niger, where greater communal divisions 
and exposure to the threat of a strengthening Boko Haram together present greater 
challenges for achieving program effectiveness. 
In both countries, factors that leave youth feeling neglected also tend to facilitate extremist 
recruitment and draw young people into violence. Participants in the qualitative data 
collection express some positive influences from the P-DEV II programs but suggest that 
programs of this sort must be expanded reach more people, and thus to combat the 
problems of frustration and limited opportunity that can undermine peace. 
V. Conclusion 
We summarize the results of the quantitative analyses in Table 51. As in previous tables, 
we show positive effects of the program, i.e., in line with P-DEV II program goals, in green, 
and negative effects, i.e., in the “wrong” direction, in red. Effects that were registered in 
the pooled cross-country analysis are shown in bold faced capital letters. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
84 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
85 
Table 51: Summary of Results 
Core vs. 
Non-core
Low Intensity 
Core vs. Non￾core
High Intensity 
Core vs. Non￾core
Panel 
Analysis
Social Cohesion
Interpersonal Trust 
Chad 
Niger 
Burkina
Institutional Trust Burkina
Community Decision￾Making Burkina Burkina Chad
Political Participation Burkina Niger 
Interethnic Marriage Chad Chad 
POOLED Burkina 
Resilience to Extremism
Access to Jobs 
Chad 
Burkina
POOLED
Access to Vocational 
Schools 
Chad 
Burkina Burkina Chad Burkina
Political Efficacy Burkina Burkina 
Perceived Ethnic 
Differences Burkina Burkina Niger
Perceived Religious 
Differences Burkina Burkina Niger
Justifiability of 
Religious Violence Burkina POOLED Burkina
Violence is Effective Chad 
Burkina Niger 
Justifiability of Violence 
in the Name of Islam Burkina Niger 
U.S. is at war with 
Islam 
Chad 
Niger 
Burkina 
POOLED
Chad 
Niger 
Burkina 
POOLED
Chad 
Niger 
Burkina 
Niger
Youth Outlook
Life Satisfaction Niger
Economic Outlook Burkina
Chad 
Niger 
Burkina
Burkina 
Inclusiveness of Middle 
Schools Chad Chad Niger 
Interest in Community 
Affairs 
Chad 
Niger 
Burkina
Political Knowledge 
Niger 
Burkina 
POOLED
Niger 
Burkina 
Niger 
Burkina 
POOLED 
Niger 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
86 
Key: Positive Effect (Statistically Significant); Negative Effect (Statistically Significant) 
The results may be summarized as follows: 
Of the 19 indicators analyzed, only two demonstrated a statistically significant pooled DiD 
effect of core zone versus non-core zone status. This means that on the overwhelming 
majority of outcomes relevant to the goals of the P-DEV II program, there was no 
detectable difference across countries in the trends over time between core and non-core 
zones. Positive pooled DiD effects were found on political knowledge (a two-item 
measure of respondent’s factual knowledge about the President’s length of term and the 
number of seats in the national legislature), and on the respondent’s perception that the 
US is not at war with Islam. All other indicators relating to, for example, trust, participation, 
access to jobs and vocational training, beliefs in the justifiability or efficacy of religious 
violence, life and economic satisfaction, and interest in community affairs, showed 
insignificant pooled DiD impacts of the P-DEV II program. 
An examination of the country-specific effects showed numerous significant impacts of 
the program, especially in Burkina Faso (see Section III-B). This may reflect the 
substantially greater number of P-DEV II activities carried out on average in Burkina core 
zones than in the two other countries. Significant DiD estimates of positive program 
impact were noted on eight indicators in Burkina Faso, five of which related to Program 
Goal 2, Resilience to Violent Extremism: access to vocational schools, political efficacy, 
reduced perceptions of ethnic and religious differences, and reduced justifiability of 
religious violence. Burkina core zones also increased more than non-core zones on 
perceptions of the inclusiveness of community decision-making, political knowledge, and 
general economic outlook, though core zones in Burkina increased more on the 
perception that the US is at war with Islam. 
The pattern of significant results suggests null findings in Chad and Niger. There were 
few significant DiD impacts in Chad or Niger (see Section III-B), with three significant 
positive impacts for core zones differences relative to non-core zones in Chad (support 
for interethnic marriage, decreased perception of the US at war with Islam, and increased 
support for gender inclusiveness in middle schools), and two in Niger (decreased 
perception of the US at war with Islam and political knowledge). 
Analysis of DiD impacts in “high intensity” zones – i.e., those with greater than average 
total P-DEV II activities – shows one additional significant pooled (cross-country) impacts 
of the program, on perceived access to jobs (see Section III-C). Differences over time on 
this indicator in the high-intensity core zones were greater than differences in the non￾core zones. Two additional indicators — the (un)justifiability of religious violence and 
support for interethnic marriage — show significant pooled effects in low-intensity zones 
compared to non-core zones, though in both cases these effects reduce to nearly zero in 
the high intensity P-DEV II zones. 
Country-by-country “high intensity” zone analysis showed a mixed pattern of impacts: 
greater numbers of indicators in Burkina were significant in the high-intensity zones, 
especially those pertaining to Resilience to Extremism. Altogether 10 of the 18 indicators 
showed significant DiD impacts in Burkina when comparing high intensity core zones to 
non-core zones. At the same time, three indicators in those Burkina zones also moved 
anomalously in the “wrong” direction, all related to Resilience to Violent Extremism. 
Similarly, in Chad and Niger there were sporadic positive impacts in high-intensity zones 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
87 
(4 in Chad 2, 2 in Niger) along with effects in the “wrong” direction (5 in Chad, 4 in Niger) 
in terms of the goals of the P-DEV II program. 
Analysis of the panel portion of the data failed to replicate the core findings of the zone￾level analyses (see Section III-D), likely due to the small number of individuals in each 
zone who were re-interviewed from the baseline wave limiting our statistical power to 
detect small programeffects. The results did show that individuals’ experiences with 
activities related to countering violent extremism increased in the core zones compared 
to individuals in the non-core zones over time. But few of those changes were associated 
positively with changes in those same individuals’ attitudes and perception since the 
baseline interviews. Both positive and negative program effects are observed in the panel 
analysis in Niger and Burkina Faso, though overall pattern of results in panel analysis in 
the two countries probably reflect essentially null effects of the program in both 
countries. All of the impacts found were relatively small in substantive magnitude, with a 
few reaching significance in the positive direction and a few in the negative 
direction. Similar balance between “near significant” effects were found in the positive 
and negative direction as well. All of this suggests that, from the panel analysis, we obtain 
a picture of generally null overall impacts in Chad and Niger. 
The study included two “list experiments” and one “endorsement” experiment designed 
to measure in unobtrusive or indirect ways the respondent’s willingness to consider 
engaging in specific acts of violent extremism or express approval of groups which 
engage in violent acts (see Section III-E). The results of these analyses suggest that there 
is a sizable pool of individuals in all country contexts who would support “an attack on [a] 
Western country’s embassy that could result in military or civilian casualties” (list 
experiment #1), who would consider “joining a group that carries out acts of violence to 
defend your religion” (list experiment #2), or who would be more likely to oppose a polio 
vaccination program if they are told that “Al-Qaeda in the Islamic Maghreb (AQIM), an 
Islamist group, will oppose this program” (endorsement experiment #1). These figures 
range from approximately 10-25% of the sampled populations. In general, however, there 
were no detectable P-DEV II core zone impacts on these outcomes, in that respondents 
in the core zones and non-core zones showed similar trends in these outcomes over time.
Examination of key informant interviews and focus groups in Burkina Faso and Niger 
showed that attitudes relating to program goals have deteriorated in the countries since 
the 2013 baseline data collection, providing context relating to support for toward violent 
extremism. This is particularly true in Niger where perceptions of program effects are 
weaker and Boko Haram has strengthened its presence. Community tensions exist in 
Burkina Faso including interreligious tensions, and struggle over land and natural 
resources. In both countries, factors that leave youth feeling neglected also tend to 
facilitate extremist recruitment and draw young people into violence. Participants in the 
qualitative data collection noted positive influences from P-DEV II programs in their 
countries but suggest that programs of this sort must be expanded reach more people, 
and thus to combat the problems of frustration and limited opportunity that can undermine 
peace. 
The overall pattern of results found in the report has several implications for future work. 
First, it is clear that the P-DEV II program did not have consistent impacts across the 
three country contexts, as there were detectable differences in the trends over time in 
pooled core versus non-core zones on only two indicators out of the 19 analyzed. Instead, 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
88 
there was significant country-by-country variation in program impact: program effects 
were strongest in Burkina Faso, where positive impacts were registered in some way 
(core versus non-core zone, high intensity core zones versus non-core) on a majority of 
the 19 indicators, with substantially fewer and weaker effects registered in Chad and 
Niger. (At the same time there were several anomalous effects in the “wrong direction” in 
all countries). This means future work will probably need to conduct more detailed 
analyses on a country-by-country basis, which will require additional numbers of 
interviews within each county in order to approach requisite levels of statistical power. 
Second, while it is possible that significant effects would be uncovered in theoretically 
meaningful subgroups of the population (e.g., men versus women, young versus old, rich 
versus poor), the fact that relatively few impacts were seen in the overall individual country 
samples – especially in Chad and Niger -- means that detecting these potential impacts 
will be extremely challenging, requiring again substantially more interviews in the endline 
data collection phase. Finally, the lack of meaningful or consistent results emerging from 
the panel portion of the analysis suggests that it may not be wise to continue these efforts 
in subsequent data collection waves, given the low payoff relative to the high cost of 
collecting data on the same individuals over time. 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
A-1 
VI. Appendix 
A. Methodology 
1. Survey Instrument 
The survey instrument developed for the P-DEV II baseline study was used as the basis 
for the midline impact evaluation. Two survey instruments, one developed by IRD￾InterMedia and the other by the EAS team serve as the foundation for the evaluation. The 
original survey was developed by IRD-InterMedia in consultation with USAID personnel, 
and served as the instrument for the first wave of interviews conducted in Chad, Niger, 
and Burkina Faso between March and September 2013. The EAS evaluation team 
designed a second survey instrument that built on the one previously designed by IRD￾InterMedia in order to improve and extend the work that had already been done. The EAS 
survey instrument kept 65% of the items in the original IRD-InterMedia survey instrument 
in order to combine information from both samples into a single data set that enhances 
the quality of the evaluation. This survey, included in the Appendix to this report, was 
altered slightly for purposes of the mid-line data collection, with approximately 25 
questions being eliminated in order to shorten interview time. 
The EAS survey instrument improved upon the original IRD-InterMedia survey in terms 
of comparability, reliability, and social desirability biases. The new survey combined the 
most powerful items from the original instrument with items that have been used in other 
comparable surveys in the region. The EAS survey instrument also takes advantage of 
new tools available to researchers for capturing the true feelings and attitudes of 
respondents faced with sensitive questions. These measures do not rely solely on 
respondent self-reports of attitudes and beliefs, and hence overcome much of the “social 
desirability” biases that can confound measures of sensitive items in the survey context, 
such as those regarding sympathy for violent and extremist groups and ideas, trust in 
local leaders, and antipathy towards women. The EAS survey instrument overcomes this 
problem through the addition of unobtrusive measures of support for violent extremism. 
Specifically, it includes a “list experiment” to capture levels of support for terrorist actions 
which may involve the deaths of civilians, as well as an “endorsement experiment” to 
measure sympathy for extremist groups operating in the Sahel region. 
From those survey items, we selected 19 indicators for analysis. In selecting and 
constructing these indicators, the EAS team was guided by recommendations in the 
current scholarly literature on violent extremism, described more fully in the baseline 
report. 
Five related to Goal 1 Social Cohesion: Interpersonal Trust, Institutional Trust, 
Perceptions of Inclusive Community Decision-Making, Political Participation, and Support 
for Interethnic Marriage. 
Nine indicators related to Goal 2, Resilience to Violent Extremism: Perceived Access to 
Jobs, Perceived Access to Vocational Schools, Political Efficacy, Perceived Ethnic 
Differences, Perceived Religious Differences, Belief in the Justifiability of Religious 
Violence, Belief that Violence is Effective, Belief in the Justifiability of Violence in the 
Name of Islam, and Belief that the U.S. is at war with Islam. 
Five indicators related to Goal 3, Youth Empowered: Life Satisfaction, Economic Outlook, 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
2 
Perceived Gender Inclusiveness of Middle Schools, Interest in Community Affairs, and 
Political Knowledge. A more detailed description of each of these indicators can be found 
in the results section (Section III-B) of this report. 
2. Survey Data Collection Interviewer Training 
Interviewer training or data collection occurred in Burkina Faso from August 6-18, 2015 
for the training of CERFODES and in Niger for August 18-30, 2015 for the training of local 
partner CESEV. EAS Team Survey and Regional Specialist Social Scientist Dr. John 
McCauley led five-day training seminars to enumerators and supervisors to explain the 
survey objectives, panel verification, experimental methodology and survey versions, and 
conduct practice sessions for administering the interviews. CERFODES technical lead 
Kadisso Nacambo was present at both of these trainings and led a follow-up training with 
additional enumerators from August 25-30 in Niger. Quality Control representatives 
Bintou Sanou (Burkina Faso), Adamou Hamadou (Niger), and Dr. Douzounet Mallaye 
(Chad) also attended the trainings, with Dr. Mallaye traveling to Niger, given security 
challenges that prevented travel to Chad at the time. Following the training, these 
provided assurance that data collection was proceeding appropriately in each country 
according to procedure. 
Since security issues prevented travel to Chad, Dr. Mallaye led an initial training from 
September 1-6, 2015. A refresher training was administered by Ms. Nacambo from March 
18-20, 2016. This refresher training was attended by Aristide Mabali, who served as 
quality control representative after delays prevented further participation from Dr. Mallaye. 
Training for qualitative data collection occurred March 3-5, 2016 in Niger and March 5-7, 
2016 in Burkina Faso. Dr. McCauley served as the lead trainer in each country, 
introducing interview and focus group methodology and data recording techniques, as 
well as targeting focus group and key informant interviews based on a preliminary 
analysis of survey data collection there occurred for fresh interviews only from March 21–
April 3, 2016. 
Procedures for “Fresh” Interviews 
The EAS Team collected 25 interviews with new randomly selected respondents in each 
zone, following random route procedures as described in the baseline report (Finkel, et 
al. 2014). This enabled analysis in zones where IRD collected baseline data as it had not 
retained information from baseline respondents to conduct panel data collection. In the 
zones where EAS collected data, the team combined observations collected in the panel 
with new observations to guard against any biases related to panel attrition or conditioning 
that may confound the analysis. 
Panel Procedures and Verification 
A portion of the data collection exercise consisted of a panel study in which individuals 
from selected communes in Niger and Burkina Faso who had been interviewed in the 
baseline survey (wave 1) were re-interviewed in the midline survey (wave 2). The sample 
consists of 500 respondents who were randomly selected from the pool of wave 1 
respondents, 25 individuals per commune for 10 communes each in Niger and Burkina 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
3 
Faso. There are four different versions of the questionnaire; meaning that in each 
commune 6 to 7 respondents received each version. EAS provided supervisors with a list 
of 25 pre-selected respondents per commune. Every household/respondent interviewed 
in the first wave of the survey was assigned an ID number. This ID number is the unique 
identifier that allows data for the same household/respondent to be known. To facilitate 
identification, the cover sheet of the questionnaire contained the address as well as any 
information available on the specific location and description of the household. 
To ensure that the exact same individuals who had been interviewed in wave 1 were re￾interviewed in the second wave, a panel verification protocol was developed which served 
as a guideline for the panel study. This panel verification protocol consists of a three￾stage verification process: the first respondent verification was made by the enumerators, 
the second verification by an EAS Quality Control Representative, and a third verification 
via random back-checks conducted by CERFODES. The protocol also entails the specific 
scenarios under which substitutions can and should be made when pre-selected 
respondents cannot be re-contacted and/or verification is unsuccessful. 
For the first stage of the verification process, enumerators were instructed to reach out to 
the exact same person interviewed in the first wave. Enumerators were asked to not 
interview another person under any circumstance. The cover sheets provided to 
enumerators contained the name of the respondent with her/his contact details (or those 
of a contact person), as well as the following information to help locate and verify the 
respondent: 1) sex, 2) age, 3) level of education, 4) nickname of respondent, and 5) name 
of the head of household. Enumerators were instructed to physically locate the selected 
respondent and confirm that the person in front of her/him is the same one that was 
interviewed in the first wave using the following verification protocol: 
1. Collect data in Section RV of the questionnaire (sex, age, level of education, 
nickname of respondent, and name of head of household). 
2. Before proceeding with the survey, verify that the person you are interviewing is 
the same person interviewed in the first wave. Cross-check the collected data from 
section RV with the information provided in the cover sheet: 
a) If there are two or more discrepancies with the information provided in the cover 
sheet, end interview and report back to supervisor. 
b) If there are two or more discrepancies with the information provided in the cover 
sheet, but to the judgment of the enumerator it is the same person interviewed 
in wave 1, the enumerator should write down detailed verification information 
justifying it and proceed with the interview. 
c) If the respondent’s name and at least four of the five characteristics correspond 
with the information provided in the cover sheet, proceed with the interview. 
For the second stage of the verification process, supervisors were instructed to provide 
information collected by enumerators (demographics and other variables) to the EAS 
Quality Control Representative in each country. These representatives cross-checked 
information from wave 1 with the information provided by the supervisors to verify that the 
person interviewed is the same one that was interviewed in the first wave. If EAS were to 
find discrepancies, it would notify the supervisors that the interview was unsuccessful and 
that the respondent should be substituted. This procedure was conducted as follows: 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
4 
1. After completing each interview, enumerators filled in the final sheet of the 
questionnaire with the following pieces of information: respondent’s name, sex, 
age, level of education, nickname, religion, ethnic background, language spoken 
at home, name of the head of household, household item: radio, and household 
item: inside or outside WC. This set of information is called the “Level 2 Verification 
sheet.” 
2. Supervisors were instructed to call the EAS Quality Control Representative at the 
end of each day of data collection and report the information on the Level 2 
Verification sheet. 
3. The EAS Quality Control Representatives were responsible for making a 
determination about whether the respondent can be verified as being the same 
person as was interviewed in wave 1. 
4. The EAS Quality Control Representative noted whether the respondent was 
verified or not at the bottom of the Level 2 Verification sheet. 
5. Unverified interviews were retained and made available to EAS. 
For the third stage of the verification process, field supervisors were instructed to conduct 
back-checks by re-contacting households/respondents that have been successfully 
interviewed as well as by calling unsuccessful interviews. More specifically, field 
supervisors were required to back-check 10% of successful re-contacts. Out of this, 100% 
were back-checked by asking 11 key questions to make sure that the right respondent 
interviewed. Similarly, field supervisors were instructed to contact 10% of unsuccessful 
interviews. If in the process of back-check, a previously unsuccessful contact was 
reached, the person doing the back-check arranged an interview with an enumerator. The 
outcome of every back-check was recorded in the QC update sheet, which was 
completed at the end of each day and sent to the project manager. 
These procedures were successful in so far as only eight respondents in the panel could 
not be verified as being the same respondent who was interviewed at baseline. These 
respondents were excluded from all analyses in the panel portion of the evaluation. 
3. P-DEV II Activity Tracker Data Collection 
In order to achieve a more fine-grained evaluation of how program implementation affects 
outcomes, the midline study utilized data on activities conducted in each commune from 
the P-DEV II Activity Tracker and Quarterly Reports prepared by IRD. The EAS Team 
collected program data from IRD recorded on a quarterly basis from the onset of the 
program. This data was then merged with the survey data aggregated at the commune 
level for further analysis. These procedures enabled the EAS Team to determine whether 
the extent of P-DEV II activities in a given zone is related to program impact, such that 
communes with more intense P-DEV II “treatments” showed larger differences in 
outcomes between baseline and midline than communes with less treatment intensity. 
This analysis augments the basic DiD analysis of program effects between core and non￾core communes over time. 
The EAS Team carried out this activity in a similar manner as it had done for preliminary 
analysis conducted and incorporated into the Appendix of the Baseline Report. The team 
aggregates all activities by commune in each country so that it can obtain measures of: 
the total number of program activities undertaken in each target zone and the type of 
USAID/WA EAS Project For The Regional Peace And Governance Programs 
Midline Report for Impact Evaluation of P-DEV II in Chad, Niger, and Burkina Faso 
5 
activities that were conducted in order to understand the nature and intensity of the 
program. 
Using the activity tracker data, the team analyzed the effect of changes in implementation 
on the changes in outcomes, measuring program intensity to estimate the effect of 
different intensities on changes in program-relevant outcomes over time. In the baseline 
report, this was conducted by comparing the program effects on zones matched by level 
of program intensity. In the midline report, this analysis was conducted with a DiD 
analysis. 
4. Qualitative Data Collection 
The midline study includes qualitative data from focus groups and interviews with key 
informants in order to provide an in-depth understanding of the context of violent 
extremism, P-DEV II experience, and attitudes relating to the program goals, in addition 
to answering questions raised by survey data analysis. 
For the baseline, nine (9) focus groups and eight (8) key informant interviews were 
conducted in Burkina; nine (9) focus groups and eight (8) key informant interviews were 
conducted in Niger; and six (6) focus groups and 15 key informant interviews were 
conducted in Chad. 
Qualitative data were collected by CERFODES in Burkina Faso and by project partner 
Cabinet d'Analyses et d'Actions pour la Sécurité et la Paix (CASPA) in Niger. Midline 
qualitative data collection included 10 focus groups and 12 interviews conducted in 
Burkina Faso and 10 focus groups and 21 interviews conducted in Niger among a 
strategic selection of actors implementing the radio, community leadership, youth 
leadership strengthening, Imam and Quranic school training, and vocational training 
activities. Given the diverse nature of the recipients, these will be complemented by 
external review of the training and strengthening activities. 
The EAS team held separate, woman-led focus groups for women to encourage 
participation and free exchange of opinion. All materials, including the interview and focus 
group field notes, were used to incorporate qualitative findings into the midline report. 
During training, Dr. McCauley and the project partners selected key zones for qualitative 
data collection based on preliminary data analysis of program effects at the community 
and individual levels. Following training, the teams initiated qualitative data collection in 
Burkina Faso on March 8 and in Niger on March 5. A list of qualitative data collection 
zones detailing focus groups and interview data collection is included in the Appendix. 
ID1 Questionnaire 
ID 
 
A-6 
B. Questionnaire 
Peace through Development II (P-DEV II) 
**EXERCISE 1 MIDLINE SURVEY [MONTH] 2015 PANEL** 
Question numbers in BLUE correspond directly to items in the original P-DEV II baseline study.
Section 0. RESPONDENT IDENTIFICATION 
Note: This section must be pre-filled before taking the survey to the field. 
ID1. Questionnaire ID: 
ID0. Questionnaire version: 
F2 Name of respondent:
O7 Sector/Quarter: 
O8 Respondent/household complete address:
O4 New localization of the respondent (for those who declared they could change the 
household): 
G4 Contact information for someone who 
could be able to help to find the 
respondent: 
ID1 Questionnaire 
ID 
 
A-7 
O9 Respondent mobile number: 
ID2. Country 
Chad 1 
Niger 2 
Burkina Faso…………………………………………..3
IDW. Wave of Interview 
ID3. Province Code 
(See location codes on page 2.) 
ID4. Target Zone Code 
(See location codes on page 2.) 
ID5. Neighborhood Name 
____________________________________ 
(write in) 
ID7. Day of interview: 
Single code 
Monday 1 
Tuesday 2 
Wednesday 3 
Thursday 4 
Friday 5 
Saturday 6 
Sunday 7 
ECode Name of Wave 2 Enumerator Signature ID7b. Date of Wave 2 Interview 
(dd/mm) 
____/_____/2015 
SCode Name of Wave 2 Supervisor Signature ID8b. Date Completed Wave 2 
Check (dd/mm) 
____/_____/2015 
F6 Record interview start time using 24 hour clock ______:______ (hour:minute) 
F7 Estimated direction from start point: 
F8 Estimated distance from start point (in M): 
ID1 Questionnaire 
ID 
 
A-8 
LOCATION CODES CHAD 
Province 
Barh El Gazal 01 Kanem 04 
Batha 02 Lac Tchad 05 
Borkou 03 N’Djamena 06 
Target Zones 
Am Djamena Bilala 01 Ngouri 07 
Assinet 02 Nokou 08 
Baga-Sola 03 Ntiona 09 
Chadra 04 Oum-Hadjer 10 
Doum-Doum 05 Wadjigui 11 
Melea 06 Yao 12 
Commune 1 N’Djamena 13 Am Sileb 15 
Am Doback 14 Commune 7 N’Dajamena 16 
LOCATION CODES NIGER 
Province 
Agadez 01 Tahoua 05 
Diffa 02 Tillaberi 06 
Maradi 03 Zinder 07 
Niamey 04
Target Zones 
Ballayara 01 Niamey 4 07 
Bosso 02 Matameye 08 
ID1 Questionnaire 
ID 
 
A-9 
G. Roumdji 03 Tébaram 09 
Iféreouane 04 Tessaoua 10 
Mayahi 05 Zinder I 11 
N'guigmi 06 
LOCATION CODES BURKINA FASO 
Province 
Kadiogo 01 Soum 04 
Oudalan 02 Yatenga 05 
Seno 03 Zondoma 06 
Target Zones 
Boussou 01 Oursi 07 
Gorgadji 02 Séguénéga 08 
Markoye 03 Seytenga 09 
Ouagadougou Arrondissement 4 04 Tongomayel 10 
Ouahigouya 05 Tougo 11 
Responses from Wave 1 Questionnaire for cross checking and verification
A1_W1 Sex of respondent Male 1 
Female 2 
A2_W1 Age of respondent years old 
A3_W1 Respondent’s educational level 
O5 Respondent's Nick name as known in 
his/her neighborhood 
O11 Name of head of household 
ID1 Questionnaire 
ID 
 
A-10 
OPENING TEXT 
Good day. My name is ____________. I am from CERFODES, an independent 
research organization, and I’m here to speak with [INSERT NAME OF RESPONDENT]. 
Is he/she available? We conducted an interview here two years ago, and he/she told 
me it would be okay to return at a later date for a follow-up. 
If you are happy to proceed, then let’s begin. 
NOTE: DO NOT BEGIN THE INTERVIEW UNLESS THE RESPONDENT GIVES 
HIS/HER CONSENT. IF THE RESPONDENT REFUSES, LEAVE THE HOUSE AND 
RECORD THE REFUSAL. 
If consent is granted: 
Thank-you. As a reminder, please know that your answers will be confidential. They will 
be put together with the responses of over 1000 other people we are talking to, to get 
an overall picture. It will be impossible to pick you out from what you say, so please feel 
free to tell us what you think. This interview will take about forty minutes. 
As we go through the questions, please keep in mind that you may say ‘I don’t know’ 
to any question when you do not know the answer, and you may say ‘pass’ to any 
question when you do not feel comfortable answering. You may also terminate the 
interview at any time. Do you understand? 
[Proceed with interview only if answer is positive]. 
Let’s begin. 
Section RV. Respondent Verification 
Q# Question Response Code Skip 
Logic 
A1 Sex of Respondent 
Do not ask. Observe. 
Single Code 
1…..Male 
2…..Female 
 
A2 How old are you? 
Below 
ID1 Questionnaire 
ID 
 
A-11 
WAIT FOR 
RESPONSE 
WRITE IN NUMBER. 
(If age not known, ask 
respondents to 
estimate in years, ask 
them whether they 
were born before or 
after an important and 
widely recognized 
event, such as a 
drought. After the 
interview, or before, if 
you know you are in 
an area where 
respondents are 
unlikely to know their 
age, establish with 
local informants the 
names of prominent 
droughts and the 
approximate years 
they took place. 
Estimate the 
respondent’s age.) 
________________ years old 
15 or 
more 
than 68 
years 
old 
STOP 
THE 
INTER
VIEW 
 
A3 What is the highest 
level of school that 
you have completed? 
1…illiterate/none 
2…no formal schooling 
3…primary incomplete 
4…primary complete 
5…secondary incomplete 
6…secondary complete 
7…University/Poly incomplete OND 
8…University/Poly complete HND 
ID1 Questionnaire 
ID 
 
A-12 
9…Post University incomplete 
10…Post University complete 
 
O5 Respondent's Nick 
name as known in 
his/her neighborhood 
 
O11 Name of head of 
household 
 
RESPONDENT VERIFICATION 
INTERVIEWER: Before proceeding with the survey, verify that the person you 
are interviewing is the same person interviewed in the first wave. Cross-check 
the data from Section RV with the information provided in the cover sheet: 
1. If there are 2 or more discrepancies with the information provided in the 
cover sheet, record code 1, end survey, and report back to supervisor. 
2. If there are 2 or more discrepancies with the information provided in the 
cover sheet but to your judgement this is the same person interviewed in 
wave 1, record code 2, write down detailed verification, and proceed with 
the survey. 
3. If the respondent’s name and at least 4 demographic characteristics 
correspond with the information provided in the cover sheet, record code 
3 and proceed with the survey. 
RV1 What can you say 
about the 
respondent? 
1… two (2) or more answers from Section RV 
do not correspond with the information 
provided in the cover sheet. I cannot verify 
that this person is the same one interviewed 
in the first wave of the survey. 
2… two (2) or more answers from Section RV 
do not correspond with the information 
provided in the cover sheet, but I can verify 
that this person is the same one interviewed 
in the first wave of the survey. 
3… four (4) or more answers from Section 
If code 
1 end 
survey 
and 
report 
to 
supervi
sor; If 
code 2 
go to 
RV2; 
If code 
ID1 Questionnaire 
ID 
 
A-13 
RV correspond with the information provided 
in the cover sheet. I can confirm with 
certainty that the person in front of me is the 
one interviewed in the first wave of the 
survey. 
3 go to 

Section 
A. 
RV2 Please provide 
detailed verification 
information 
Go to 
Section 
A. 
 
Section A. Demographics 
 
A4 Are you currently 
employed or 
unemployed? 
1…Employed 
2…Unemployed 
88…Don’t know 
99…Refused 
If code 
2 go to 
A5; 
Else go 
to A6 
 
A5 If you’re not working, 
what is your status? 
WAIT FOR 
RESPONSE 
1…Student 
2…Non-working pensioner or invalid 
3…Housewife/maternity leave 
4…Looking for work 
5…Not looking for work 
6…Waiting for work to start 
7…Other non-working, specify 
88…Don’t know 
99…Refused 
 
ID1 Questionnaire 
ID 
 
A-14 
A6 I’m going to read a list 
of items and 
amenities that you 
may have inside or 
around your house. 
Please tell me 
whether you have or 
do not have each 
item. 
READ OUT ITEMS. 
CIRCLE ALL THAT 
APPLY 
a. Fridge/freezer 
b. Computer or iPad 
c. Video or DVD player 
d. Satellite dish 
e. TV 
f. Radio 
g. Telephone (land) 
h. Telephone (mobile) 
i. Air conditioning 
j. Washing machine 
k. Car 
l. Gas or electric cooker 
m. Inside or outside WC 
n. Inside or outside pipe borne tap 
 
A7 Speaking now about 
religion, which religion 
are you? 
AWAIT REPLY. 
Circle ONLY ONE 
religious group 
Muslim ................................................................ 1 
Shia .................................................................... 2 
Ismaili ................................................................. 3 
Izala/Wahabit ...................................................... 4 
Sunni Muslim .................................................... 5 
Maliki ................................................................. 6 
Hanafi ................................................................. 7 
Shafi’I ................................................................. 8 
Kadria ................................................................. 9 
Hanbali ............................................................. 10 
Tidjaniya ........................................................... 11 
Christian ........................................................... 12 
Catholic ............................................................ 13 
Protestant ......................................................... 14 
Orthodox .......................................................... 15 
Other Christian ................................................ 16 
Jewish ............................................................... 17 
Buddhist ............................................................ 18 
ID1 Questionnaire 
ID 
 
A-15 
Hindu ................................................................ 19 
Traditional beliefs ........................................... 20 
Baha’i ............................................................... 21 
I don’t belong to any religion ........................... 22 
Other…. ......................................................... 23a 
If other, specify:………………………………23b 
Don’t know 88 
Refused ............................................................. 99 
 
A8 How important is 
religion in your 
personal life? Is it very 
important, somewhat 
important, or not very 
important? 
SHOW CARD 01 
1…not very important 
2…somewhat important 
3…very important 
88…Don’t know 
99…Refused 
 
A9 How often do you 
attend religious 
services? 
READ OUT 
REPONSES. 
SINGLE CODE. 
1… Several times a day 
2… Once a day 
3… Several times per week 
4…Once a week 
5… Once a month 
6… Only for religious holidays or special occasions 
7… Rarely or never 
88…Don’t know 
99…Refused 
 
A10 How often do you 1… Several times a day 
ID1 Questionnaire 
ID 
 
A-16 
pray? 
WAIT FOR 
RESPONSE 
SINGLE CODE 
2… Once a day 
3… Several times per week 
4…Once a week 
5… Once a month 
6… Only for religious holidays or special occasions 
7… Rarely or never 
88…Don’t know 
99…Refused 
 
A11 And what is your ethnic background? 
WAIT FOR RESPONSE 
MULTIPLE CODES 
 CHAD NIGER BURKINA FASO 
Arab 1 
Kanembou 2 
Gourane 3 
Peuls 4 
Moussei 5 
Sara 6 
Zaghawa 7 
Kanuri 8 
Boudouma 9 
Moundang 10 
Massa 11 
Other, Specify 
 12 
Don’t know 88 
Arab 1 
Toubou 2 
Haussa 3 
Fulbe 4 
Zarma/Songhai 5 
Gourmantche 6 
Tuareg 7 
Kanuri 8 
Beriberi 9 
Boudouma 10 
Manga 11 
Other, Specify 12 
Don’t know 88 
Refused 99 
Mossi 1 
Fulani (Peul) 2 
Bobo/Dioula 3 
Senoufo 4 
Goumantche 5 
Lobi 6 
Gurunsi 7 
Dagaaba 8 
Tuareg 9 
Other, Specify 12 
Don’t know 88 
Refused 99 
ID1 Questionnaire 
ID 
 
A-17 
Refused 99 
 
A12 Which language do you speak most at home? 
WAIT FOR RESPONSE 
SINGLE CODE 
 CHAD NIGER BURKINA FASO 
French 1 
Arabic 2 
Kanembou 3 
Fulfulde 4 
Gourane 5 
Kanuri 6 
Sara 7 
Zaghawa 8 
Boudouma 9 
Moundang 10 
Massa 11 
Other, specify 
 12 
Don’t know 88 
Refused 99 
French 1 
Arabic 2 
Haussa 3 
Fulfulde 4 
Gourmanchema 5 
Kanuri. 6 
Zarma/Songhai 7 
Tamashek 8 
Toubou 9 
Tagdalt. 10 
Other, specify 12 
Don’t know 88 
Refused 99 
French 1 
Mooré 2 
Fulfuldé 3 
Dioula/Bambara/ 
Malinké 4 
Gourmanchema 5 
Lobiri. 6 
Gurunsi 7 
Dagaare 8 
Hausa 9 
Tagdalt. 10 
Tuareg/Berber 11 
Other, specify 12 
Don’t know 88 
Refused 99 
 
A13 How many other 
languages do you 
understand well 
enough to listen to a 
0….None 
1…One other language 
2…Two to four other languages 
ID1 Questionnaire 
ID 
 
A-18 
radio broadcast in that 
language? 
3…Five to seven other languages 
4…More than seven other languages 
 
Section B. Participation, Community Decision Making, Groups, and Trust 
Q# Question Response Code Skip 
Logic
B1 Now I would like to ask you some questions about how decisions are 
made in your community. When important decisions are made in this 
commune/neighborhood/village/camp, how much DO the following people 
participate: not at all, somewhat, or a lot? 
READ OUT RESPONSES 
USE SHOW CARD 02 
SINGLE CODE IN EACH ROW 
Not at 
all 
Somewhat A 
lot 
DK Refus
e 
a) Ordinary people from the 
commune/ neighborhood 1 2 3 88 99 
b) Youth 1 2 3 88 99 
c) Women 1 2 3 88 99 
d) People from your own tribe 
or ethnic group 1 2 3 88 99 
e) People who share your 
religious views 1 2 3 88 99 
And how much do you think the following people SHOULD participate 
ID1 Questionnaire 
ID 
 
A-19 
B2 when important decisions are made in this 
commune/neighborhood/village/camp? 
READ OUT RESPONSES 
USE SHOW CARD 02 
SINGLE CODE IN EACH ROW 
Not at 
all 
Somewhat A 
lot 
DK Refus
e 
a) Ordinary people from the 
commune/ neighborhood 1 2 3 88 99 
b) Youth 1 2 3 88 99 
c) Women 1 2 3 88 99 
d) People from your own tribe 
or ethnic group 1 2 3 88 99 
e) People who share your 
religious views 1 2 3 88 99 
B3 Are you satisfied, neutral, or 
dissatisfied with the way decisions are 
made in your community? 
READ OUT RESPONSES. 
USE SHOWCARD 02a 
SINGLE CODE 
1…Dissatisfied 
2…Neutral 
3…Satisfied 
88…DK 
99…Refused 
B4 Now I’d like your opinion on some governance issues. I’m going to read a 
list of statements, and I’d like you to please tell me whether you agree with 
the following statements: not at all, somewhat, or quite. 
ID1 Questionnaire 
ID 
 
A-20 
SHOW CARD 02a 
Not at 
all Somewhat Quit
e 
DK Refus
e 
a) My opinions are respected 
by local leaders 1 2 3 88 99 
b) Local government takes into 
account the opinions of citizens 
in decision-making processes 
1 2 3 88 99 
c) I feel well-prepared for 
participating in political life 1 2 3 88 99 
d) People like me have no say 
in what the government does 1 2 3 88 99 
 
B5 I’m going to list a number of activities. Please tell me whether you have or 
have not done each activity in the past 12 months. 
Single Code in Each Row 
Yes No DK Refuse 
a) Attended a commune/neighborhood 
councilor other public meeting 1 2 88 99 
b) Met with an elected official, called 
him/her, or sent a letter 1 2 88 99 
c) Notified the village chief about a local 
problem 1 2 88 99 
d) Voted 1 2 88 99 
e) Participated in community 
development activities 1 2 88 99 
f) Joined a protest or demonstration 1 2 88 99 
ID1 Questionnaire 
ID 
 
A-21 
B6 Speaking about your role in the community, I am going to read out a list 
of groups that people join or attend. For each one, could you tell me 
whether you are a member or not a member? 
SHOW CARD 03 
Not a 
member 
Active 
member 
Inactive 
member 
DK Refus
e 
a) A religious group (e.g. 
a mosque, church) 1 2 3 88 99 
b) A trade union or 
farmers association 1 2 3 88 99 
c) A professional or 
business association 1 2 3 88 99 
d) A community or self￾help association 1 2 3 88 99 
e) A youth group 1 2 3 88 99 
f) Some other voluntary 
association or 
community group. 
SPECIFY____________
____ 
1 2 3 88 99 
 
B7 ASK B7 IF CODE 2 HAS BEEN CHOSEN FOR AT LEAST ONE OF THE 
QUESTIONS IN B6; IF NOT GO TO B8. 
Thinking about the members of the group(s) to which you belong, do any 
of the groups… 
Yes No DK Refuse 
a) Include both men and women? 1 2 88 99 
b) Include people of different religious 
views? 1 2 88 99 
ID1 Questionnaire 
ID 
 
A-22 
c) Include people of different ethnic 
groups? 1 2 88 99 
B8 I will now read you two statements about your neighborhood. Please tell 
me if you agree or disagree with these statements. 
SHOW CARD 04 
Disagre
e 
Neither 
agree nor 
disagree 
Agree DK Refus
e 
a) Most people are 
willing to help if you ask 
for help 
1 2 3 88 99 
b) It is naïve to trust 
people 1 2 3 88 99 
B9 Now I would like to ask you about the trust you have in different groups of 
people. Please tell me if you agree or disagree with the following 
statements. 
READ OUT RESPONSES. 
 
USE SHOWCARD 04. 
Disagre
e 
Neither 
agree nor 
disagree 
Agree DK Refuse 
a) I trust local authorities 1 2 3 88 99 
b) I trust central 
government 1 2 3 88 99 
c) I trust religious leaders 1 2 3 88 99 
ID1 Questionnaire 
ID 
 
A-23 
d) I trust non￾governmental 
organizations 
1 2 3 88 99 
e) I trust the police 1 2 3 88 99 
g) I trust the military 1 2 3 88 99 
B10 On how many days out of the last 7 did you… 
Number of days DK Refus
e 
a) talk to someone who lives in a 
different community 88 99 
b) talk to someone with different 
religious view 88 99 
c) talk to someone from a different ethnic 
group from yours 88 99 
B10B How strongly do you feel you belong to 
your immediate 
community/neighborhood? 
1…Very strongly 
2…Fairly strongly 
3…Not very strongly 
4…Not at all strongly 
88…DK 
99…Refused 
B11 In elections, Nigeriens/Chadians/ Burkinabe often vote for candidates 
from their own ethnic group. Which of the following statements is closer to 
your view? 
READ OUT RESPONSES 
SHOW CARD 05 
Circle ONLY ONE CODE 
ID1 Questionnaire 
ID 
 
A-24 
1. It is normal to want to elect 
someone from your ethnic 
community 
2. Voters should place much less 
emphasis on ethnic 
considerations. 
1…Selects statement 1 
2…Selects statement 2 
88…DK 
99…Refused 
B12 Do you agree or disagree with the 
following statement: 
I tell my children (or I will tell my future 
children) that they should only marry 
people from their same ethnic group 
READ OUT RESPONSES 
SHOW CARD 04 
1…Disagree 
2…Neither agree nor disagree 
3…Agree 
88…DK 
99…Refused 
Section C. Life Satisfaction, Political Interest, Media Use, Economic and Services 
Evaluation 
Q# Question Response Code Skip 
Logic
C1 Here is a ladder representing the “ladder of life.” Let’s suppose the top of 
the ladder (10) is the best possible life for you; and the bottom (0), the 
worst possible life for you. On which step of the ladder do you personally 
stand at the present time? 
USE SHOWCARD 06. 
SINGLE CODE. 
Worst Life …..………………………………………… Best Life 
0 1 2 3 4 5 6 7 8 9 10 
Don’t know…88 
Refused… 99 
ID1 Questionnaire 
ID 
 
A-25 
C3 And on which step do you think you will stand in the future, say two years 
from now? 
 
USE SHOWCARD 06. 
SINGLE CODE. 
Worst Life …..………………………………………… Best Life 
0 1 2 3 4 5 6 7 8 9 10 
Don’t know…88 
Refused… 99 
C4 Thinking first of your community, would 
you say you have a great deal of 
interest, some interest, or very little 
interest in local community affairs? 
READ OUT RESPONSES 
SHOW CARD 07 
Circle ONLY ONE CODE 
1…very little interest 
2…some interest 
3…a great deal of interest 
88…DK 
99…refused 
C6 In general, would you say that the 
economy of [COUNTRY] is better, 
worse, or about the same than it was a 
year ago? 
READ OUT RESPONSES 
SHOW CARD 08 
Circle ONLY ONE CODE 
1…worse 
2…the same 
3…better 
88…DK 
99…refuse 
C6
A 
In general, would you say that your 
household financial situation is better, 
1…worse 
ID1 Questionnaire 
ID 
 
A-26 
worse, or about the same than it was a 
year ago? 
READ OUT RESPONSES 
SHOW CARD 08 
Circle ONLY ONE CODE 
2…the same 
3…better 
88…DK 
99…refuse 
C7 In general, would you say that the 
country is headed in the right direction 
or the wrong direction? 
READ OUT RESPONSES 
SHOW CARD 09 
Circle ONLY ONE CODE 
1…right direction 
2…wrong direction 
88…DK 
99…refused 
C8 Now I would like to ask you about services that are available in your 
community. Are you satisfied or dissatisfied with your access to… 
READ OUT RESPONSES 
SHOW CARD 10 
Circle ONLY ONE CODE IN EACH ROW 
Satisfied Dissatisfied DK Refus
e 
a) Education/Schools 1 2 88 99 
b) Health services/clinics/ 
hospitals 1 2 88 99 
c) Access to water, 
electricity, and other 
services 
1 2 88 99 
d) Government legal 
services (courts) 1 2 88 99 
ID1 Questionnaire 
ID 
 
A-27 
C10 Now I would like to ask you some questions specifically about middle 
school and vocational school 
Please tell me if you agree or disagree with the following statements... 
READ OUT RESPONSES. 
USE SHOWCARD 04. 
SINGLE CODE IN EACH ROW. 
 Disagree Neither Agree DK Ref. 
 a) Education is only 
for boys, but not for 
girls 
1 2 3 88 99 
 b) Vocational 
training is accessible 
for people like me 
1 2 3 88 99 
C11 How difficult is it to get a job in 
[COUNTRY] today: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
Circle ONLY ONE CODE 
1…Not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Refused 
C13 Now I would like to ask you a few 
questions about how politics works 
in [YOUR COUNTRY]. Do you 
know how long the term of office is 
for the President of [Country]? That 
is, after he is elected, how many 
years does he stay in office before 
the next election? 
[All countries = 5 years] 
1…correct answer given
2…incorrect answer given 
88….DK 
99…Ref 
ID1 Questionnaire 
ID 
 
A-28 
C14 Do you happen to know how many 
seats there are in the National 
Assembly? 
[Burkina Faso = 127] 
[Chad = 188] 
[Niger = 113] 
Number given by respondent 
[ ] 
88…DK 
99…Ref 
C15 How well would you say the government is managing the following 
matters? 
Very badly Fairly Very 
well 
DK Ref 
a) Improving living 
standards 1 2 3 88 99 
b) Keeping the community 
safe 1 2 3 88 99 
c) Improving public services 1 2 3 88 99 
d) Dealing with violent 
groups in the region 1 2 3 88 99 
C16 Now, I would like to ask you about rights of citizens. 
In your view, how well does the government protect the following rights: 
Not at all Somewh
at 
Very 
well 
DK Ref 
a) Freedom of expression 
and opinion 1 2 3 88 99 
b) Protection from arbitrary 
arrest and ensuring fair 
trials 
1 2 3 88 99 
c) Protection from torture 1 2 3 88 99 
ID1 Questionnaire 
ID 
 
A-29 
and cruel, inhuman or 
degrading punishment 
d) Equal treatment before 
the law 1 2 3 88 99 
C17 Please tell me how much you agree that women should be equal to men 
regarding the following matters. 
Disagre
e 
strongly
Somewh
at 
Disagree 
Somewh
at Agree 
Agree 
strongly
DK R
ef 
a) In family and 
household 
matters 
1 2 3 4 88 99 
b)With regards to 
owning property 
and receiving 
inheritance 
1 2 3 4 88 99 
c) With regards to 
employment 1 2 3 4 88 99 
d) Being a 
candidate in local, 
parliamentary, 
and presidential 
elections 
1 2 3 4 88 99 
e) Being a clan 
chief 1 2 3 4 88 99 
C18 In your opinion, how often do 
Nigerien/Chadian/Burkinabe 
government officials who commit 
crimes [break the law] get punished? 
1…Always 
2…Rarely 
3…Never 
88…DK 
99…Refuse 
ID1 Questionnaire 
ID 
 
A-30 
C19 In your opinion, over the past year, 
has the level of corruption in 
Niger/Chad/Burkina Faso increased, 
decreased, or stayed the same? 
1…Increased 
2…Stayed the same 
3…Decreased 
88…DK 
99…Refuse 
Section D. Violence and Extremism 
Q# Question Response Code Skip 
Logic 
D1A Now I would like to ask you about your 
community. Differences often exist 
between people living in the same 
village/neighborhood. To what extent do 
you feel that ethnic differences tend to 
divide people in your 
village/neighborhood? 
READ OUT RESPONSES 
SHOW CARD 02B 
1…Not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Refuse 
D1B Do ethnic divisions in your 
village/neighborhood ever lead to 
violence? 
1…yes 
2…no 
88…DK 
99…Refuse 
D2A To what extent do you feel that religious 
differences tend to divide people in your 
village/neighborhood? 
READ OUT RESPONSES 
SHOW CARD 02B 
1…Not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Refuse 
ID1 Questionnaire 
ID 
 
A-31 
D2B Do religious divisions in your 
village/neighborhood ever lead to 
violence? 
1…yes 
2…no 
88…DK 
99…Refuse 
D2C How often have you or members of your 
family ever been unfairly treated because 
of your ethnic background? 
1…Often 
2…Sometimes 
3…Never 
88…DK 
99…Refused 
D2D How often have you or members of your 
family ever been unfairly treated because 
of your religious views? 
1…Often 
2…Sometimes 
3…Never 
88…DK 
99…Refused 
D2E In the past 12 months, have you or has 
someone else in your household been 
physically attacked? 
1…Yes, me. 
2…Yes, someone in my 
household. 
3…Yes, both me and 
someone in my household. 
4…No 
88…DK 
99…Refused 
D3 When do you think that violence is an 
effective method to solve problems: 
1…Often 
2…Sometimes 
ID1 Questionnaire 
ID 
 
A-32 
often, sometimes, or never? 
READ OUT RESPONSES 
SHOW CARD 11 
3…Never 
88…DK 
99…Refused 
D4 Some people think that using arms and 
violence against civilians in defense of 
their religion is justified. Other people 
believe that, no matter what the reason, 
this kind of violence is never justified. Do 
you personally feel that using arms and 
violence against civilians in defense of 
your religion can be often justified, 
sometimes justified, or never justified? 
READ OUT RESPONSES 
SHOW CARD 11 
1…Often 
2…Sometimes 
3…Never 
88…DK 
99…Refused 
D6 Please tell me whether you agree or disagree with the following 
statements 
READ OUT RESPONSES 
SHOW CARD 04 
Circle ONLY ONE CODE 
Disagre
e 
Neither Agree DK Ref 
a) Violence in the name of 
Islam can be justified 1 2 3 88 99 
b) The United States is at war 
against Islam, not terrorism 1 2 3 88 99 
d) The government should 
work with western countries to 
fight terrorism 
1 2 3 88 99 
e) Al Qaeda’s violent actions 
are permitted under Islamic 
law 
1 2 3 88 99 
ID1 Questionnaire 
ID 
 
A-33 
D7 Which of these three statements comes closest to your own opinion? 
READ OUT RESPONSES 
SHOW CARD 12 
Circle ONLY ONE CODE 
a) Democracy is preferable to any other 
form of government 
b) In certain situations, a nondemocratic 
form of government can be preferable 
c) It doesn’t matter to me what form of 
government we have 
1…Selects A 
2…Selects B 
3…Selects C 
88…DK 
99…REF 
D8 Some people say we would be better off if the country was governed 
differently. What do you think about the following options? Do you agree 
or disagree that… 
READ OUT RESPONSES 
SHOW CARD 04 
Circle ONLY ONE CODE IN EACH ROW 
Disagree Neither Agree DK Ref 
a) we should get rid 
of elections so that a 
strong leader can 
decide everything 
1 2 3 88 99 
b) We should be 
governed by Sharia 
Law 
1 2 3 88 99 
c) we should let the 
military rule the 
country 
1 2 3 88 99 
d) we should only 
have one political 
1 2 3 88 99 
ID1 Questionnaire 
ID 
 
A-34 
party 
D9A Now, I would like to ask you again about your community. 
In the last 12 months, has there been violence in your community 
associated with: 
If code 1 for 
any go to 
D9B 
Else go to 
Yes No DK Ref D10 
a1) A strike or protest 1 2 88 99 
b1) Conflict over land 1 2 88 99 
c1) Another type of conflict 1 2 88 99 
D9B Do you think that the violence 
associated with these events was 
justified? 
1…Yes 
2…No 
88…DK 
99…Refused 
Respondents will be split for question D10. Half of the respondents will 
respond to D10: Control Group, and the other half will respond to D10: 
Treatment Group. 
D10: Control CONTROL GROUP ONLY 
If respondent answers to D10 Control, skip directly to D11. 
 
 
Imagine that you hear on the radio that a 
Western newspaper has published 
offensive images of the Prophet 
Mohammed committing a crime. You 
could respond to this affront in many 
ways. I’m going to read you a list of 
possible responses now. Please listen to 
them and then tell me how many of the 
following reactions you would support. 
Please, do not tell me which options are 
a. A peaceful protest 
taking place at the 
Western country's 
embassy 
b. Your government 
demanding a formal 
apology from the 
Western country 
c. Your government 
declaring war against 
the Western country 
ID1 Questionnaire 
ID 
 
A-35 
necessary, only how many. 
READ OUT RESPONSES 
SHOW CARD 14 
WRITE THE NUMBER OF OPTIONS 
DK 88 
Refuse 99
D10: Treatment TREATMENT GROUP ONLY 
Only ask item below if respondent did not receive control 
group item 
 Imagine that you hear on the radio that a 
Western newspaper has published 
offensive images of the Prophet 
Mohammed committing a crime. You 
could respond to this affront in many 
ways. I’m going to read you a list of 
possible responses now. Please listen to 
them and then tell me how many of the 
following reactions you would support. 
Please, do not tell me which options are 
necessary, only how many. 
READ OUT RESPONSES 
SHOW CARD 15 
WRITE THE NUMBER OF OPTIONS 
a. A peaceful protest 
taking place at the 
Western country's 
embassy 
b. Your government 
demanding a formal 
apology from the 
Western country 
c. Your government 
declaring war against 
the Western country 
d. An attack on the 
country’s embassy 
that could result in 
military or civilian 
casualties 
DK 88 
Refuse 99
Respondents will be split for question D11. Half of the respondents will 
respond to D11: Control Group, and the other half will respond to D11: 
Treatment Group. 
D11: Control CONTROL GROUP ONLY 
If respondent answers D11 Control, skip directly to D12. 
 The World Health Organization recently 
announced a plan to introduce universal 
polio vaccination across {Country}. How 
much do you approve of such a plan? 
READ OUT RESPONSES 
1…not at all 
2…somewhat 
3…quite 
88…DK 
ID1 Questionnaire 
ID 
 
A-36 
SHOW CARD 02a 
CIRCLE ONE CODE 
99…Refuse 
D11: Treatment TREATMENT GROUP ONLY 
Only ask item below if respondent did not receive control group 
item 
 The World Health Organization recently 
announced a plan to introduce universal 
polio vaccination across {Country}. It is 
likely that Al-Qaeda in the Islamic 
Maghreb (AQIM)/Boko Haram, an 
Islamist group, will oppose this program. 
How much do you approve of such a 
plan? 
READ OUT RESPONSES 
SHOW CARD 02a 
CIRCLE ONE CODE 
1…not at all 
2…somewhat 
3…quite 
88…DK 
99…Refuse 
Respondents will be split for question D13. Half of the respondents will 
respond to D13: Control Group, and the other half will respond to D13: 
Treatment Group. 
D13: Control CONTROL GROUP ONLY 
If respondent answers D13 Control, skip directly to Section E.
OK. Now to change the subject, I am 
going to read you a list of things that 
people in this area may have considered 
doing. Please listen to them and then 
tell me how many YOU would consider 
doing. 
Please, do not tell me which ones, only 
how many. 
READ OUT RESPONSES 
SHOW CARD D13C 
a. Donating clothes or 
money to someone in 
need 
b. Voting for a 
candidate who is a 
strong supporter of 
homosexual rights 
c. Reporting an official 
who has taken bribes
d. Going on the Hajj 
88…DK 
99…Refuse 
ID1 Questionnaire 
ID 
 
A-37 
WRITE THE NUMBER OF OPTIONS 
D13: Treatment TREATMENT GROUP ONLY 
Only ask item below if respondent did not receive control 
group item. 
 OK. Now to change the subject, I am 
going to read you a list of things that 
people in this area may have considered 
doing. Please listen to them and then 
tell me how many YOU would consider 
doing. 
Please, do not tell me which ones, only 
how many. 
READ OUT RESPONSES 
SHOW CARD D13T 
WRITE THE NUMBER OF OPTIONS 
a. Donating clothes or 
money to someone in 
need 
b. Voting for a 
candidate who is a 
strong supporter of 
homosexual rights 
c. Reporting an official 
who has taken bribes
d. Going on the Hajj 
e. Joining a group that 
carries out acts of 
violence to defend 
your religion 
88…DK 
99…Refuse 
Section E. Peace and Tolerance Radio Module 
Q# Question Response Code Skip 
Logic 
E1 Generally speaking, how often 
have you heard messages or 
conversations about peace and 
tolerance? 
READ OUT RESPONSES 
SHOW CARD 11 
CIRCLE ONE CODE 
1…Often 
2…Sometimes 
3…Never 
88…DK 
99…Ref 
E2 How often within the last month have you… 
ID1 Questionnaire 
ID 
 
A-38 
READ OUT RESPONSES 
SHOW CARD 17 
CIRCLE ONE CODE IN EACH ROW 
Rarel
y 
Once 
a 
mont
h 
Several 
times a 
week
Ever
y day
D
K 
Ref
a) watched television 1 2 3 4 88 99 
b) read a newspaper 1 2 3 4 88 99 
c) used the internet 1 2 3 4 88 99 
d) made or received voice 
calls on a mobile phone 
1 2 3 4 88 99 
e) sent or received an SMS 
(text message) on a mobile 
phone 
1 2 3 4 88 99 
f) listened to the radio 1 2 3 4 88 99 
E3 What do you usually listen to on 
the radio? 
READ OUT RESPONSES. 
CIRCLE ALL THAT APPLY. 
a) Music 
b) News 
c) Drama 
d) Religious Program 
e) Youth Programs 
f) Governance Programs 
g) Peace and Tolerance 
programs 
h) Other (specify) 
i) DK 
j) Ref 
CHAD 
Q# Question Response Code Skip 
Logic 
E4a Have you ever listened to the radio 
program Dabalaye (the meeting place)? 
1…Yes 
2…No If code 
ID1 Questionnaire 
ID 
 
A-39 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT:
It’s a governance program with 
presenters Mahamouth and Allamine 
88…DK 
99…Ref 
2, 88 
or 99 
go to 
E11a 
E5a 
IF YES AT E4a: How often would you say 
you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
1… Several times a week 
2… Once a week 
3… Once a month 
4… Less than once a month
88…DK 
99…Ref 
E6a How much do you like [Dabalaye]: not at 
all, somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E7a How much do you trust the information 
you hear on [Dabalaye]: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8a How many people have you spoken to 
about things you heard on the radio show 
[Dabalaye]? No one, one or two people, 
three or four people, or more than four 
people? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
ID1 Questionnaire 
ID 
 
A-40 
E10a Generally, when you listen to [Dabalaye], 
can you tell me who you are with when 
you listen? Alone, with some friends, with 
family, in a formal listening club? 
1… alone 
2… with some friends 
3… with family 
4… in a formal listening club 
88… DK 
99… Ref 
E11a How many people have spoken to you 
about things they heard on the radio show 
[Dabalaye]? Yes (how many times) or 
No? 
READ OUT RESPONSES 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
Go to 
E4b 
Now I would like to talk about the radio program Chabab Al Haye
E4b Have you ever listened to the radio 
program Chabab Al Haye (Youth Alive)? 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT: 
It’s a youth program with presenters 
Djamila and Abdelfatha 
1…Yes 
2…No 
88…DK 
99…Ref 
If 
code 
2, 88 
or 99 
go to 
E11b 
E5b 
IF YES AT E4b: How often would you 
say you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
1… Several times a week 
2… Once a week 
3… Once a month 
4… Less than once a month
88…DK 
99…Ref 
E6b How much do you like [Chabab Al Haye]: 
not at all, somewhat, or a lot? 
1…not at all 
2…Somewhat 
3…A lot 
ID1 Questionnaire 
ID 
 
A-41 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
88…DK 
99…Ref 
E7b How much do you trust the information 
you hear on [Chabab Al Haye]: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8b How many people have you spoken to 
about things you heard on the radio show 
[Chabab Al Haye] in the last year? no 
one, one or two people, three or four 
people, or more than four people ? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
E10b Generally, when you listen to [Chabab Al 
Haye], can you tell me who you are with 
when you listen? Alone, with some 
friends, with family, in a formal listening 
club? 
1… alone 
2… with some friends 
3… with family 
4… in a formal listening club 
88… DK 
99… Ref 
E11b 
Has anybody spoken to you about things 
they heard on the radio show [Chabab Al 
Haye]? Yes (how many times) or No? 
READ OUT RESPONSES 
1… Yes, several times 
2… Yes, a few times 
3… No 
88…DK 
99…Ref 
Go to 
Sectio
n F 
ID1 Questionnaire 
ID 
 
A-42 
NIGER 
Q# Question Response Code Skip 
Logic 
E4a Have you ever listened to the radio 
program Sada Zumunci (Solidarity)? 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT: 
It’s a governance and religious affairs 
program with presenters Mairo and 
Abdou 
1…Yes 
2…No 
88…DK 
99…Ref 
If 
code 
2, 88 
or 99 
go to 
E11a 
E5a 
IF YES AT E4a: How often would you say 
you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
1… Several times a week 
2… Once a week 
3… Once a month 
4… Every three months
88…DK 
99…Ref 
E6a How much do you like [Sada Zumunci]: 
not at all, somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E7a How much do you trust the information 
you hear on [Sada Zumunci]: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8a 1…no one 
ID1 Questionnaire 
ID 
 
A-43 
How many people have you spoken to 
about things you heard on the radio show 
[Sada Zumunci] in the last year? No one, 
one or two people, three or four people, 
or more than four people? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
E10a Generally, when you listen to [Sada 
Zumunci], can you tell me who you are 
with when you listen? Alone, with some 
friends, with family, in a formal listening 
club? 
1… alone 
2… with some friends 
3… with family 
4… in a formal listening club 
88… DK 
99… Ref 
Go to 
E4b 
E11a 
Has anybody spoken to you about things 
they heard on the radio show [Sada 
Zumunci]? Yes (how many times) or No?
READ OUT RESPONSES 
1… Yes, several times 
2… Yes, a few times 
3… No 
88…DK 
99…Ref 
Go to 
E4b 
Now I would like to talk about the radio program Gwadaben Matassa (Youth Boulevard)
E4b Have you ever listened to the radio 
program Gwadaben Matassa (Youth 
Boulevard)? 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT: 
It’s a youth program with presenters 
Djamila and Abdelfatha 
1…Yes 
2…No 
88…DK 
99…Ref 
If 
code 
2, 88 
or 99 
go to 
E11b 
E5b 1… Several times a week 
ID1 Questionnaire 
ID 
 
A-44 
IF YES AT E4b: How often would you 
say you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
2… Once a week 
3… Once a month 
4… Less than once a month
88…DK 
99…Ref 
E6b How much do you like [Gwadaben 
Matassa]: not at all, somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E7b How much do you trust the information 
you hear on [Gwadaben Matassa]: not at 
all, somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8b 
How many people have you spoken to 
about things you heard on the radio show 
[Gwadaben Matassa] in the last year? no 
one, one or two people, three or four 
people, or more than four people? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
E10b Generally, when you listen to [Gwadaben 
Matassa], can you tell me who you are 
with when you listen? Alone, with some 
friends, with family, in a formal listening 
1… alone 
2… with some friends 
3… with family 
Go to 
Sectio
n F 
ID1 Questionnaire 
ID 
 
A-45 
club? 4… in a formal listening club 
88… DK 
99… Ref 
E11a 
Has anybody spoken to you about things 
they heard on the radio show [Gwadaben 
Matassa]? Yes (how many times) or No? 
READ OUT RESPONSES 
1… Yes, several times 
2… Yes, a few times 
3… No 
88…DK 
99…Ref 
Go to 
Sectio
n F 
BURKINA FASO 
Q# Question Response Code Skip 
Logic
E4a Have you ever listened to the radio 
program Malegr Sooré (Voices of 
Change)? 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT: 
It’s a youth program in Mooré with 
presenters Angèle and Eric. 
1…Yes 
2…No 
88…DK 
99…Ref 
If 
code 
2, 88 
or 99 
go to 
E11a
E5a 
IF YES AT E4a: How often would you say 
you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
1… Several times a week 
2… Once a week 
3… Once a month 
4… Less than once a month
88…DK 
99…Ref 
E6a How much do you like [Malegr Sooré]: not 
at all, somewhat, or a lot? 
READ OUT RESPONSES 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
ID1 Questionnaire 
ID 
 
A-46 
SHOW CARD 02 
SINGLE CODE 
99…Ref 
E7a How much do you trust the information 
you hear on [Malegr Sooré]: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8a 
How many people have you spoken to 
about things you heard on the radio show 
[Malegr Sooré] in the last year? No one, 
one or two people, three or four people, 
or more than four people? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
88…DK 
99…Ref 
E10a Generally, when you listen to [Malegr 
Sooré], can you tell me who you are with 
when you listen? Alone, with some 
friends, with family, in a formal listening 
club? 
1… alone 
2… with some friends 
3… with family 
4… in a formal listening club 
88… DK 
99… Ref 
Go to 
E4b 
E11a 
Has anybody spoken to you about things 
they heard on the radio show [Malegr 
Sooré]? Yes (how many times) or No? 
READ OUT RESPONSES 
1… Yes, several times 
2… Yes, a few times 
3… No 
88…DK 
99…Ref 
Go to 
E4b 
Now I would like to talk about the radio program Pinal Sukabè (Youth Awakening)
ID1 Questionnaire 
ID 
 
A-47 
E4b Have you ever listened to the radio 
program Pinal Sukabè (Youth 
Awakening)? 
IF RESPONDENTS SAY THEY DON’T 
KNOW THE PROGRAM, READ OUT: 
It’s a youth program in Mooré with 
presenters Issouf and Kady. 
1…Yes 
2…No 
88…DK 
99…Ref 
If 
code 
2, 88 
or 99 
go to 
E11b 
E5b 
IF YES AT E4b: How often would you 
say you listen to the program? 
USE SHOWCARD 18. 
SINGLE CODE. 
1… Several times a week 
2… Once a week 
3… Once a month 
4… Less than once a month
88…DK 
99…Ref 
E6b How much do you like [Pinal Sukabè]: not 
at all, somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E7b How much do you trust the information 
you hear on [Pinal Sukabè]: not at all, 
somewhat, or a lot? 
READ OUT RESPONSES 
SHOW CARD 02 
SINGLE CODE 
1…not at all 
2…Somewhat 
3…A lot 
88…DK 
99…Ref 
E8b 
How many people have you spoken to 
about things you heard on the radio show 
[Pinal Sukabè] in the last year? No one, 
one or two people, three or four people, 
1…no one 
2…one or two people 
3…three or four people 
4…more than four people 
ID1 Questionnaire 
ID 
 
A-48 
or more than four people? 
READ OUT RESPONSES 
SHOW CARD 19 
SINGLE CODE 
88…DK 
99…Ref 
E10b Generally, when you listen to [Pinal 
Sukabè], can you tell me who you are with 
when you listen? Alone, with some 
friends, with family, in a formal listening 
club? 
1… alone 
2… with some friends 
3… with family 
4… in a formal listening club 
88… DK 
99… Ref 
Go to 
Sectio
n F 
E11b [Ask only if E4b is 2, 88, or 99] 
Has anybody spoken to you about things 
they heard on the radio show [Pinal 
Sukabè]? Yes (how many times) or No? 
READ OUT RESPONSES 
1… Yes, several times 
2… Yes, a few times 
3… No 
88…DK 
99…Ref 
Go to 
Sectio
n F 
Section F. Experience with P-DEV II Programming 
F3 Please tell me how many times over the past year you have attended any of the 
following kinds of events or activities: 
Read out Responses 
Use Show Card 20 
Single Code in Each Row 
ID1 Questionnaire 
ID 
 
A-49 
Never One or 
two 
times 
Three 
or four 
times 
More 
than 
four 
times 
DK Refus
e 
a) An organized workshop 
or forum in your community 
where issues about peace 
or democracy were 
discussed 
1 2 3 4 88 99 
b) A poetry reading or a 
drama about peace or 
democracy issues 
1 2 3 4 88 99 
c) Training to develop ways 
to solve community 
problems 
1 2 3 4 88 99 
d) An organized workshop 
or forum at a place other 
than a mosque where 
imams or religious leaders 
discussed peace or 
democracy with people in 
the community 
1 2 3 4 88 99 
e) Trainings to help you find 
work or find a better job 1 2 3 4 88 99 
f) Trainings to help you 
learn how to run a business 
or write an application for a 
business development 
grant 
1 2 3 4 88 99 
g) A meeting to help 
determine how community 
grant funding will be spent 
1 2 3 4 88 99 
Section O: Panel Follow-up Instructions and Respondent Location 
ID1 Questionnaire 
ID 
 
A-50 
 Thank you for taking the time to answer our survey. 
We may return in 2 years to ask you some more questions. I hope that I will be the 
person who will return, but it may be another person in my place. Because I may 
not be the same person, may I ask you a few more questions just to make sure we 
find you again? 
G2 Would you be willing to speak with 
us again in the next few years to we 
can see how you’re doing? 
Yes…………………………1 
No………………………….2 
G4 Could you give us the contact 
information for someone who you 
expect will be able to help us find 
you in the next few years? 
[WRITE NAME AND PHONE NUMBER OF 
CONTACT] 
O1 Can we find you the same time of day 
during the week? 
Yes…………………………1 
No………………………….2 
O2 Which is the best time to reach you? 
O3 Can we find you again in the next two 
years here? 
Yes…………………………1 
No………………………….2 
O4 If No, where can we find you and what is the best way to get there? 
Interviewer: note the precise location and instructions for finding the house for the 
next wave: 
O6 Respondent's 
occupation/professio
n 
ID1 Questionnaire 
ID 
 
A-51 
O7 Sector/ Quarter 
O8 Respondent/househo
ld complete address : 
O9 Respondent mobile 
number : 
O10 Respondent position 
in the household 
Section G: Interviewer Data Entry 
G5 Record interview end time using 24 
hour clock 
_____:_____ (hour:minute) 
FOR INTERVIEWER COMPLETION ONLY --DO NOT ASK RESPONDENT! 
PLEASE COMPLETE THIS SECTION AS SOON AS POSSIBLE AFTER YOU 
HAVE LEFT THE RESPONDENT'S HOME.
G6 Length of interview ____________ Minutes 
 
G7 Sex of interviewer Male………………………………….....
1 
Female…………………………..………
2 
 
G8 In what language was the interview 
conducted? 
French 1 
Arabic 2 
French and Arabic 3 
ID1 Questionnaire 
ID 
 
A-52 
Kanembu 4 
Haussa 5 
Zarma 6 
Tamashek 7 
Gourane 8 
Other, specify ___________________
 9 
 
G9 
[Afr 
Bar] 
Were there any other people 
immediately present who were 
listening during the interview? 
No 
one.......................................................
.1 
Spouse 
only......................................................
.2 
Children 
only..................................................3 
A few 
others...............................................4 
Small 
crowd...............................................5 
 
G10 
[AfrB
ar] 
What was the respondent’s attitude toward you during the interview? 
A. Was he or she Friendly 
1 
In between 
2 
Hostile 
3 
B. What he or she Interested 
1 
In between 
2 
Bored 
3 
C. Was he or she Cooperative 
1 
In between 
2 
Uncooperative 
3 
D. Was he or she Patient 
1 
In between 
2 
Impatient 
3 
ID1 Questionnaire 
ID 
 
A-53 
E. Was he or she At ease 
1 
In between 
2 
Suspicious 
3 
F. Was he or she Honest 
1 
In between 
2 
Misleading 
3 
 
G11 
[Afr 
Bar] 
What proportion of the questions do 
you feel the respondent had difficulty 
answering? 
All 
............................................................
1 
Most 
.........................................................2 
Some 
........................................................3 
Few 
..........................................................4 
None 
........................................................5 
 
G12 On which attempt was the interview 
completed? 
First 
attempt................................................
1 
Second 
attempt............................................2 
Third 
attempt...............................................3 
 
G13 Please note any questions that caused particular difficulties for the respondent: 
[CONTRACTOR: Do not code this list for the data set. Only provide 
summary of problem questions in the Technical Report]. 
____________________________________________________________
________________ 
_______________________________________________________________
ID1 Questionnaire 
ID 
 
A-54 
_______________________________________________________________
__________________________ 
ID1 Questionnaire 
ID 
 
A-55 
Peace through Development II (PDEVII) 
LEVEL 2 Verification Sheet 
***REPORT TO SUPERVISOR ONLY*** 
PLEASE TABULATE IN FRENCH 
Responses from Wave 2 Questionnaire for cross checking and verification
ID1 Questionnaire 
ID 
F2 Name of 
Respondent 
A1_W1 Sex of 
respondent 
Male 1 
Female 2 
A2_W1 Age of 
respondent 
 years old 
A3_W1
Respondent’s 
educational 
level 
O5 
Respondent's 
Nick name as 
known in his/her 
neighborhood 
O11 Name of head 
of household 
A3 Education Level
A7 Religion 
A11 Ethnic 
Background 
A6F Household 
Item: Radio Non…0 Yes…1 
A6M Household WC Inside or Outside 
ID1 Questionnaire 
ID 
 
A-56 
Item: Inside or 
Outside WC 
NOTE FOR EAS TEAM QUALITY CONTROL REPRESENTATIVE ONLY: 
RV3 : Verified……… 1 
Not Verified ….2 
 
A-57 
B. Regression Tables for Panel Analysis 
Table 52. Panel Analysis for Interpersonal Trust 
Variable Name Coefficient 
Niger dummy -.262 
differenced no. of activities .003 
differenced no. of activities*Niger dummy .004 
lagged dependent variable -1.148*** 
lagged dependent variable * Niger dummy .210** 
constant 2.538*** 
* p < .10; ** p < .05; *** p < .01 
Table 53. Panel Analysis for Institutional Trust 
Variable Name Coefficient 
Niger dummy -.161 
differenced no. of activities -.011* 
differenced no. of activities*Niger dummy .005 
lagged dependent variable -.872*** 
lagged dependent variable * Niger dummy .034 
constant 2.306 
* p < .10; ** p < .05; *** p < .01 
 
A-58 
Table 54. Panel Analysis for Community Decision-making 
Variable Name Coefficient 
Niger dummy -.226 
differenced no. of activities -.006 
differenced no. of activities*Niger dummy .009 
lagged dependent variable -.965*** 
lagged dependent variable * Niger dummy .068 
constant 2.059 
* p < .10; ** p < .05; *** p < .01 
Table 55. Panel Analysis for Political Participation 
Variable Name Coefficient 
Niger dummy .045 
differenced no. of activities .001 
differenced no. of activities*Niger dummy .019** 
lagged dependent variable -.851*** 
lagged dependent variable * Niger dummy .024 
constant .311*** 
* p < .10; ** p < .05; *** p < .01 
 
A-59 
Table 56. Panel Analysis for Interethnic Marriage 
Variable Name Coefficient 
Niger dummy .235 
differenced no. of activities -.005 
differenced no. of activities*Niger dummy .002 
lagged dependent variable -.914*** 
lagged dependent variable * Niger dummy -.075 
constant 2.612*** 
* p < .10; ** p < .05; *** p < .01 
Table 57. Panel Analysis for Access to Jobs 
Variable Name Coefficient 
Niger dummy .495*** 
differenced no. of activities .005 
differenced no. of activities*Niger dummy -.006 
lagged dependent variable -1.000*** 
lagged dependent variable * Niger dummy -.064 
constant 1.235*** 
* p < .10; ** p < .05; *** p < .01 
 
A-60 
Table 58. Panel Analysis for Access to Vocational School 
Variable Name Coefficient 
Niger dummy .506** 
differenced no. of activities .034** 
differenced no. of activities*Niger dummy -.031 
lagged dependent variable -.889*** 
lagged dependent variable * Niger dummy -.261** 
constant 1.792 
* p < .10; ** p < .05; *** p < .01 
Table 59. Panel Analysis for Political Efficacy 
Variable Name Coefficient 
Niger dummy -.118 
differenced no. of activities -.002 
differenced no. of activities*Niger dummy .008 
lagged dependent variable -.927*** 
lagged dependent variable * Niger dummy -.059 
constant 1.789 
* p < .10; ** p < .05; *** p < .01 
 
A-61 
Table 60. Panel Analysis for Ethnic Differences 
Variable Name Coefficient 
Niger dummy .519*** 
differenced no. of activities .005 
differenced no. of activities*Niger dummy .033 
lagged dependent variable -.908*** 
lagged dependent variable * Niger dummy .008 
constant 1.097*** 
* p < .10; ** p < .05; *** p < .01 
Table 61. Panel Analysis for Religious Differences 
Variable Name Coefficient 
Niger dummy .460** 
differenced no. of activities .007 
differenced no. of activities*Niger dummy .035 
lagged dependent variable -.922*** 
lagged dependent variable * Niger dummy .088 
constant 1.111*** 
* p < .10; ** p < .05; *** p < .01 
 
A-62 
Table 62. Panel Analysis for Religious Violence Is Justified 
Variable Name Coefficient 
Niger dummy .303*** 
differenced no. of activities -.001 
differenced no. of activities*Niger dummy -.002 
lagged dependent variable -.975*** 
lagged dependent variable * Niger dummy -.030 
constant .994*** 
* p < .10; ** p < .05; *** p < .01 
Table 63. Panel Analysis for Violence Is Effective to Solve Problems 
Variable Name Coefficient 
Niger dummy .233 
differenced no. of activities .007 
differenced no. of activities*Niger dummy -.034** 
lagged dependent variable -.899*** 
lagged dependent variable * Niger dummy -.150 
constant 1.155*** 
* p < .10; ** p < .05; *** p < .01 
 
A-63 
Table 64. Panel Analysis for Violence in the Name of Islam 
Variable Name Coefficient 
Niger dummy -.097 
differenced no. of activities .011* 
differenced no. of activities*Niger dummy -.004 
lagged dependent variable -.978*** 
lagged dependent variable * Niger dummy .047 
constant 1.113*** 
* p < .10; ** p < .05; *** p < .01 
Table 65. Panel Analysis for Anti-West Attitudes 
Variable Name Coefficient 
Niger dummy .592** 
differenced no. of activities .022** 
differenced no. of activities*Niger dummy -.007 
lagged dependent variable -.993*** 
lagged dependent variable * Niger dummy -.095 
constant 1.377*** 
* p < .10; ** p < .05; *** p < .01 
 
A-64 
Table 66. Panel Analysis for Life Satisfaction 
Variable Name Coefficient 
Niger dummy 1.164*** 
differenced no. of activities .044 
differenced no. of activities*Niger dummy -.040 
lagged dependent variable -.998*** 
lagged dependent variable * Niger dummy -.148* 
constant 4.346*** 
* p < .10; ** p < .05; *** p < .01 
Table 67. Panel Analysis for Economic Outlook 
Variable Name Coefficient 
Niger dummy .013 
differenced no. of activities -.025** 
differenced no. of activities*Niger dummy .023 
lagged dependent variable -.819*** 
lagged dependent variable * Niger dummy .054 
constant 1.573*** 
* p < .10; ** p < .05; *** p < .01 
 
A-65 
Table 68. Panel Analysis for Inclusiveness of Schools 
Variable Name Coefficient 
Niger dummy .152 
differenced no. of activities .001 
differenced no. of activities*Niger dummy -.015 
lagged dependent variable -.920*** 
lagged dependent variable * Niger dummy -.111 
constant 2.661*** 
* p < .10; ** p < .05; *** p < .01 
Table 69. Panel Analysis for Interest in Community Affairs 
Variable Name Coefficient 
Niger dummy -.140 
differenced no. of activities .007 
differenced no. of activities*Niger dummy -.012 
lagged dependent variable -.868*** 
lagged dependent variable * Niger dummy .002 
constant 2.208*** 
* p < .10; ** p < .05; *** p < .01 
 
A-66 
Table 70. Panel Analysis for Political Knowledge 
Variable Name Coefficient 
Niger dummy .077* 
differenced no. of activities .004 
differenced no. of activities*Niger dummy .005 
lagged dependent variable -.774*** 
lagged dependent variable * Niger dummy -.016 
constant .321*** 
* p < .10; ** p < .05; *** p < .01 
 
A-67 
C. Qualitative Data Collection Focus Group and Interviews 
Burkina Faso 
Data Collection 
Zones Core/Noncore Data Collection 
Dori Core Interview: 
 Radio Commentator 
Gorgadji Noncore Focus Group: Men 
 Interviews: 
 Female Director of Community Action Committee 
 Catholic Priest 
 Ouagadougou 
Arrondissement 9 Core Focus Group: Women, urban, educated 
Focus Group: Young Men, urban, educated 
 Ministry Official 
 Parliamentary Assistant 
Ouahigoya Core Focus Group: Urban Women 
 Interviews: 
 Civil Society Organization Leader 
 Radio Commentator 
 Imam 
 Youth Leader 
Ouahigoya Core Focus Group: Men 
Markoye Core Focus Group: rural men 
 Interview: 
 Administrative Authority 
Namissiguima (Yatenga) Noncore Focus Group: Muslim men 
 Interviews: 
 Administrative Authority 
Séguénéga Non-Core Focus Group: Men 
Seytenga Core Focus Group: Women Muslims 
Zondoma Focus Group: Women, rural, non-educated 
 
A-68 
Niger 
Data Collection Zones Core/Noncore Data Collection 
Agadez Core Focus Group: Women 
 Interviews: 
 Civil Society Activist 
 Local Authority 
Arlit Core Focus Group: Youth 
 Interviews: 
 Civil Society Activist 
 Local Elected Official 
 Religious Leader 
Balleyara Core Focus Group: Women 
 Interviews: 
 Local Elected Official 
Diffa Core Focus Group: Men 
 Interviews: 
 Civil Society Activist 
 Traditional Leader 
 Religious Leaders 
Maradi Core Focus Group: Men 
 Interviews: 
 Civil Society Activist 
 Religious Leader 
 Local Elected Official 
Niamey Gamkallé Core Focus Group: Women 
 Interviews: 
 Local Elected Official 
 President of Women's Society 
 Youth Association 
Niamey Talladjé Core Focus Group: Men 
Tahoua Core Focus Group: Men 
 Interviews: 
 Religious Leader 
 Civil Society Activist 
Tillaberi Core Focus Group: Youth 
 Interviews: 
 Local Elected Official 
 Religious Leader 
Zinder Core Focus Group: Youth 
 Interviews: 
 
Local Elected Official 
Religious Leader (Pastor) 
 
A-69 
 Religious Leader (Pastor) 
U.S. Agency for International Development 
1300 Pennsylvania Avenue, NW 
Washington, DC 20523 
Tel: (202) 712-0000 
Fax: (202) 216-3524 
www.usaid.gov