Although vignettes featuring hypothetical politicians are popular in survey experiments, political scientists rarely explore the tradeoffs between realism and experimental control. In this manuscript, I argue for greater use of “mirror experiments” in political science, or controlled survey experiments that use real politicians and mimic real world situations. This article demonstrates that cognitive burden, brought on by the lack of familiarity with a hypothetical politician, can bias experimental results. Without prior knowledge of a politician, attitudes toward political actors are artificially malleable in laboratory settings, exaggerating treatment effects on traditional approval questions. This article concludes with two empirical demonstrations of these issues, and provides a set of criteria for judging when realistic or hypothetical survey vignettes should be used.

No abstract

No abstract

No abstract

No abstract