Fig 1.
Treatment effects on knowledge.
Estimates of the treatment effect of WhatsApp messages on a standardized index of responses to factual questions that relate to the messages sent. 95% confidence intervals plotted. All specifications include week fixed effects. Standard errors clustered at the week-broadcast list level.
Fig 2.
Treatment effects on behavior.
Estimates of the treatment effect of WhatsApp messages on behavior measured through a list experiment between subscribers in treated and control broadcast lists. 95% confidence intervals plotted. All specifications include week fixed effects. Standard errors clustered at the week-broadcast list level.
Fig 3.
Subgroup treatment effects on knowledge.
Estimates of the treatment effect of WhatsApp messages on a standardized index of responses to factual questions that relate to the messages sent. 95% confidence intervals plotted. All specifications include randomization block fixed effects and (apart from by-week estimates) week fixed effects. Standard errors clustered at the week-broadcast list level.
Fig 4.
Subgroup treatment effects on behavior.
Estimates of the treatment effect of WhatsApp messages on behavior measured through a list experiment between subscribers in treated and control broadcast lists. 95% confidence intervals plotted. All specifications include randomization block fixed effects and (apart from by-week estimates) week fixed effects. Standard errors clustered at the week-broadcast list level.
Fig 5.
Trusted sources of information about COVID-19.
Respondents were asked to select up to three sources of information that they trust most on WhatsApp to deliver information about COVID-19.