Abstract
Social science experiments often expose participants to false, deceptive, or otherwise harmful content. In an effort to mitigate the effects of such content and to comply with regulatory standards, these studies usually conclude by "debriefing" participants about the content they encountered, on the assumption that doing so will eliminate the effects of exposure. We present evidence showing that this assumption is not always correct. After standard debriefs, participants who have seen political misinformation often remain worse-informed than if they had never been exposed to misinformation in the first place. We then design and test a new approach to debriefing, which entails (i) debriefing as soon after exposure to harmful content as possible; (ii) providing an informative correction; and (iii) requiring participants to affirm that they have been exposed to false information. Across multiple experiments, we show that this approach is far superior to standard debriefs at reducing the effects of false or harmful content on political attitudes and beliefs. Our approach makes it possible to study the effects of false or harmful content on attitudes and beliefs without posing significant risks to participants.