Transparency-sunshine-is often touted as a core element of the governance agenda, and one that is most important in environments with low transparency to begin with. In a provocative paper published in the American Political Science Review, Edmund Malesky, Paul Schuler, and Anh Tran present the results of a creative experiment in which they provided an additional spotlight on the activities of a random sample of delegates to Vietnam's National Assembly. They report that the effect of sunshine was negative, that delegates subject to this treatment curtailed their speech, and that those who spoke most critically were punished through the subsequent election and promotion processes. The present paper argues that Malesky, Schuler, and Tran's results, if interpreted correctly, actually predict a net positive effect of transparency. The differences in interpretation stem primarily from three sources: the interpretation of regression results for models with interaction terms, the interpretation of the variable for Internet penetration, and significant pre-treatment differences between treated and control delegates. For the context in which more than 80 percent of delegates operate, Malesky, Schuler, and Tran's results predict a positive but insignificant effect of transparency. In addition, Internet penetration, itself a measure of access to information, is positively associated with critical speech. The paper draws lessons for the design and interpretation of randomized experiments with interaction effects
