Male Librarians Cause More Confusion than Female Engineers: Evidence on the Difficulties in Correctly Assessing the Gender Composition of Occupations from a Survey Experiment

Abstract

This paper reports an unexpected result from a survey experiment that examined whether womenand men in gender-atypical occupations are perceived as less desirable dating partners than thosein gender-typical occupations. In our preregistered study, we asked participants from an onlineopt-in panel to evaluate profiles of potential dating partners in a total of 60 male- or female-dominated occupations. Contrary to expectations, we found no effect of the vignette person’soccupation on perceived dating desirability. When exploring the potential reasons for our nullfinding, we found that participants did not always interpret occupational gender compositionas intended. While participants generally understood gender-typing in low-status roles, theyoften misidentified medium- and high-status female-dominated occupations—particularly whenmale vignette profiles were presented in these roles. These misperceptions attenuated the aver-age treatment effect. Design-specific features of our experiment—such as the use of multipleoccupations per treatment category and partial confounding between occupation and vignettegender—likely further weakened the treatment signal. Rather than interpreting this case as asimple experimental “failure,” we use it to illustrate how survey experiments may fall short ofthe intended identification strategy when respondents interpret experimental stimuli differentlythan researchers anticipate and what precautions may be derived from our study.