While acknowledged by the authors as more diverse than college students, an Amazon Mechanical Turk (MTurk) sample may not be fully representative of the broader American electorate, potentially limiting the generalizability of the findings on voter behavior.
The use of a single gender-neutral hypothetical candidate, 'Jamie Easton,' might not fully capture the complexities of voter reactions to real-world candidates with established reputations, specific public personas, or varying types of actual allegations. This could limit the ecological validity of the experiment.
Simplified Allegation Scenario
The specific wording of the sexual harassment allegation (groping, harassment, and a settled lawsuit) was chosen to be 'not extremely violent' and a 'relatively minor legal issue.' This simplification might not reflect the diverse range and severity of real-world allegations, potentially leading to an underestimation or altered perception of voter response to more serious or unproven claims.
Self-Reported Voting Intention
The dependent variable relies on self-reported 'likelihood to vote,' which is a stated intention and may not perfectly translate to actual voting behavior in a real election. Social desirability bias could also influence responses.
Blunt Proxies for Underlying Values
The authors acknowledge that using Democrat/Republican affiliation as a proxy for social liberalism/conservatism and 'rape myth acceptance' is a blunt measure. More fine-grained psychological measures of these underlying values could provide a deeper understanding of the causal mechanisms at play.