Data and code for "Digital Information Provision and Behavior Change: Lessons from Six Experiments in East Africa"

While some studies suggest mobile phone-based information programs change behavior, others find no effect. We evaluate six text-message agricultural extension programs, collectively covering 128,000 farmers. A meta-analysis finds a 1.22-fold increase in the odds of adoption of recommended practices (95% CI: 1.16,1.29). We cannot reject similar impacts across experiments. Impacts are increased by message repetition, but not by providing more granular information, using behavioral framings, or complementing texts with phone calls. There is little evidence of message fatigue or crowd-out. Despite modest absolute impacts detectable only with large samples or a meta-analysis, texts are inexpensive enough to be cost-effective. The accompanying files provide guidance, data, and code to replicate the analysis.