close× Call Us
Title The Effects of Media Messages on Social Attitudes in Uganda
Post date 01/19/2016
C1 Background and Explanation of Rationale

The objective of this study is to estimate the impact of short video vignettes on the attitudes of viewers and those in their local community. Specifically, the project evaluates the impact of dramatized messages on attitudes towards three issues: stigma against women who have had abortions, domestic violence against women, and teacher absenteeism. Three short videos for each issue were produced and screened in rural Uganda, where these issues are prominent social concerns. The video vignettes were inserted in commercial breaks during films that were presented free of charge during a four-week film festival during November-December of 2015. Our experimental sample contains fifty-six village clusters that were assigned to seven different treatment arms: 3 pairs of messages, 3 single-message treatments, and one pure control. Outcomes will be measured on the individual level through a household survey of 40 adults in each village cluster. A baseline survey has been conducted in half of the village clusters.

C2 What are the hypotheses to be tested?

The overarching hypothesis is that viewers will become sensitized to the norms dramatized in the video vignettes and diffuse them throughout the community. Our initial analysis focuses on what might be termed intent-to-treat effects, i.e., the effect of village-level assignment regardless of whether the respondent attended the film festival. The ITT summarizes the effect of all forms of treatment influence: direct exposure, exposure via friends or family who attended the festival, and discussions within the community. This analysis addresses the policy question of whether providing public service announcements on topics such as these has effects on villagers’ survey responses weeks after the intervention. Additionally, our analysis will focus on three important subgroups: The first group may be called Compliers – those who would attend a film festival regardless of the video messages embedded in its commercial breaks. A second subgroup comprises Indirect Compliers: those who did not attend the film(s) themselves but report that family or friends attended. The residual group are those who, according to their endline survey reports, neither attended nor have direct family or friendship ties to those who attended. We hypothesize that treatment effects become steadily weaker as we move from Compliers to Indirect Compliers to Others. Finally, we also test for the effect of receiving a baseline survey in one’s village where we hypothesize that the baseline survey itself will heighten sensitivity to social norms, which will amplify the effect of the media treatments. Please see the attached pre-analysis plan for more details.

C3 How will these hypotheses be tested? *

Outcomes will be measured using a large number of individual survey questions concerning abortion, domestic violence, and teacher absenteeism to be administered to villagers during our endline survey. To keep the analysis manageable and to improve the reliability of our outcome measures, additive indexes will be created for two aspects of each issue domain: personal views and perceived social norms. To test our main hypotheses, we will use Seemingly Unrelated Regression (SUR) models where we first aggregate our data to the village cluster level in order to take into account that subjects were cluster assigned by village. Additionally, we will we employ a hierarchical linear modeling strategy. The model accounts for clustering at the village cluster level through a varying trade center (cluster) level intercept, which is associated with its own set of predictors taken from the covariates used in the SUR specifications. Individual outcomes are associated with the treatment indicator, a block indicator, and other individual covariates. Please see the attached pre-analysis plan for more details on the specific models that will be run.

C4 Country Uganda
C5 Scale (# of Units) Baseline: 1107 respondents in 28 clusters. Endline: 2240 respondents in 56 clusters. 82 interviews with village leaders.
C6 Was a power analysis conducted prior to data collection? Yes
C7 Has this research received Insitutional Review Board (IRB) or ethics committee approval? Yes
C8 IRB Number AAAP6500
C9 Date of IRB Approval 08/14/2015
C10 Will the intervention be implemented by the researcher or a third party? Innovations for Poverty Action (IPA)
C11 Did any of the research team receive remuneration from the implementing agency for taking part in this research? No
C12 If relevant, is there an advance agreement with the implementation group that all results can be published? Yes
C13 JEL Classification(s) not provided by authors