Designing Scenario-Based Experiments: Does the Use of Third-Person Framing of Vignettes and Questions Reduce Social Desirability Bias?

Abstract

Much behavioral accounting research investigates the antecedents of morally questionable behavior, such as managers' earnings management and auditors' acceptance of aggressive reporting by clients. This research often employs scenarios. It has been recommended that these scenarios and the questions that participants answer about these scenarios be framed in the third person instead of the second person to prevent social desirability bias, and many accounting researchers follow this recommendation. However, empirical evidence that third-person framing debiases responses to morally sensitive scenarios is rare. We use a pre-registered experiment with 1,121 participants to examine if the framing of the scenario itself (the vignette) and the framing of the question to measure the dependent variable, independently or in conjunction, affect social desirability bias in participants' responses. Our results provide no evidence that this is the case, raising doubts about whether researchers in accounting and related fields are really better off using third-person frames than second-person frames.

Keywords: social desirability bias, vignette framing, question framing, moral sensitivity