Call Us: US - +1 845 478 5244 | UK - +44 20 7193 7850 | AUS - +61 2 8005 4826

Sanctioning past business performance

To improve precision of estimating treatment effects, I limit the
analysis to areas of civic activity I would expect treatment to impact. Blind to treatment
group, civic events were coded into one of twelve categories: grievances, public goods, town
hall meetings, taxes, political party, associations, personal conflicts, NGO activity, administrative, religious/traditional, and for-profit. Contrary to the other categories, I would not
expect the treatment to have an effect on events categorized as religious10 (496 events) or forprofit11 (1491 events) in nature. Not only did the civics course explicitly state that religious
activities fall outside the civic domain, but Bleck (2013) finds that religious and political
participation are inversely related in Mali.
The category NGO activity (1419 events, 343 of which refer to participation in treatment)
helps address another measurement problem: the lack of a placebo condition. The control
group received no intervention when ideally they would have received a course on something
other than civic education. If individuals have a time budget for engaging in NGO activities,
then their participation in the course itself is likely to decrease participation in other NGOsponsored events. For instance, a participant in the civics course may be just as likely to
attend a meeting convened by the chief, but much less likely to attend the next NGO project
meeting. This would bias against finding a treatment effect.