Validating a willingness to share measure of a vignette experiment using real-world behavioral data

Willingness measures are widely employed in vignette and conjoint experiments. However, the predictive validity of these measures relies on the degree to which expressed intentions correspond with behavior in the real world. This study uses a unique research design that enables to validate an intention measure. In a 2022 online survey, respondents expressed their willingness to donate their social media data in hypothetical scenarios. Nine months later the same respondents were invited to participate in a genuine data donation study. The results show that the correlation between willingness and actual donation was moderate (η: 0.28). Additionally, the determinants of self-reports and sharing behavior did not overlap. The willingness to donate was more strongly associated with softer attitudinal variables, whereas actual data sharing was linked more closely to harder demographic variables. These findings challenge the predictive validity of willingness to share measures and highlight the need for realistic scenarios in survey experiments, minimizing participants' perceived risks of the task in question, and addressing social desirability bias.