Our mixed-methods study—combining experimental, quasi-experimental, descriptive, and qualitative methods—assesses four of the five RTP components. Specifically, the components covered by this study are (1) Strengthening the Inspectorate Services of the RNP, (2) Strengthening the Rule of Law for Policy Reform, (3) Media Strengthening, and (4) Strengthening Civic Participation. Collectively, these four components account for over 85 percent of the RTP’s total program implementation budget. To determine the scope of the evaluation, we first sought to identify implemented activities within a component that could potentially produce a measurable effect within the study’s timeframe. For example, in coordination with MCC, we decided not to study effects of the Media Strengthening component’s assistance to media associations, because only a small part of the original activity plan for these associations was implemented. Next, we examined each activity to determine whether it would be possible to obtain data that would generate meaningful evidence on the program’s effects. We sought to collect information about either a comparison group similar to those who received the program or the prior status of program beneficiaries before activities began. Several RTP activities did not meet this requirement. For example, the RNP Strengthening and Media Strengthening components provided training and technical assistance to a large number of journalists and RNP officers before our evaluation was initiated, precluding the collection of baseline data or the identification of a comparison group. Likewise, activities that were nationwide in scope, such as the legislative policy reform initiatives supported by the Strengthening Rule of Law component or efforts to support passage of a national media reform law as part of the Media Strengthening component, were not included in the evaluation due to the absence of baseline data or an identifiable comparison group. It is possible that the RTP may have had indirect, longer-term impacts on national policies and institutions that fall beyond the scope of our evaluation designs.
Kind of Data
Sample survey data [ssd]
Unit of Analysis
Anonymized dataset for public distribution
Program implementers, CSO leaders, local government officials, and RNP officers
Producers and sponsors
Mathematica Policy Research
Millennium Challenge Corporation
The baseline and follow-up surveys each had a target sample size of 10,000 respondents. Each survey included a different sample of respondents (i.e. data is cross-sectional, not longitudinal). To ensure that the sample was representative and widely distributed across the country, sample targets were calculated at the sector level. Using the most recent national census, we calculated the proportion of the national population within each sector. We determined the number of individuals to survey in each sector by applying that proportion to our targeted sample size of 10,000. The survey sample included all 416 sectors in Rwanda. Within each sector, households were selected using a random walk method, and one adult respondent (age 16 or older) was selected at random within each household.
Dates of Data Collection
Data Collection Notes
The baseline and follow-up surveys each had a target sample size of 10,000 respondents. Each survey included a different sample of respondents (i.e. data is cross-sectional, not longitudinal). The baseline survey had a target sample size of 10,000 respondents. To ensure that the sample was representative and widely distributed across the country, sample targets were calculated at the sector level. Using the most recent national census, we calculated the proportion of the national population within each sector and . We determined the number of individuals to survey in each sector by applying that proportion to our targeted sample size of 10,000. The survey sample included all 416 sectors in Rwanda. Within each sector, households were selected using a random walk method, and one adult respondent (age 16 or older) was selected at random within each household.
Roddom Consult Ltd.
Citizen Survey Questionnaire Design: the citizen survey questionnaire focused on activities implemented under three components of the RTP: RNP Strengthening, Media Strengthening, and Strengthening Civic Participation.
We used a data cleaning process designed to resolve inconsistencies in survey responses, survey question skip-patterns, and out-of-range data. These cleaning measures were implemented via SAS statistical software (version 9). To create a final data file for analysis and public use, we consulted with MCC regarding guidelines to ensure that all respondent personally identifiable information (PII) would be protected, primarily by destroying records of respondent names and by grouping outlier survey responses with potentially identifying information (such as very high age or income levels) into a uniform upper limit, or “top code.”
Data entry was completed using CSPro software.
Estimates of Sampling Error
The standard errors of the impact estimates were corrected for the possibility of correlations among individuals' outcomes within sectors.
Monitoring & Evaluation Division
Millennium Challenge Corporation
Rwanda Threshold Program Evaluation Citizen Survey, Mathematica Policy Research (2011 and 2012).
Version 1.0 (October 2014)
Version 2.0 (May 2015). Edited version based on Version 01 (DDC-MCC-RWA-THRESHOLD-MPR-2014-v1.1) that was done by Millennium Challenge Corporation.