Are the effects of informational interventions driven by salience?

Informational interventions have been shown to significantly change behavior across a variety of settings. Is that because they lead subjects to merely update beliefs in the right direction? Or, alternatively, is it to a large extent because they increase the salience of the decision they target, affecting behavior even in the absence of inputs for belief updating? We study this question in the context of an informational intervention with school parents in Brazil. We randomly assign parents to either an information group, who receives text messages with weekly data on their child's attendance and school effort, or a salience group, who receives messages that try to redirect their attention without child-specific information. We find that information makes parents more accurate about student attendance, and has large impacts on their test scores and grade promotion relative to the control group. Even though salience messages, in contrast, do not make parents more accurate about attendance levels, learning outcomes in the salience group improve by at least as much. Why? We show that treated parents across both conditions become more accurate about changes in their children's grades over time, although not about grade levels. Such coarse belief updating is consistent with independent information acquisition in response to salience effects from both interventions. Our results have implications for the design and interpretation of informational interventions across a range of domains.