Measuring norms: Assessing the threat of social desirability bias to the Bicchieri and Xiao elicitation method

Bicchieri and Xiao (2009) pioneered a method for eliciting normative expectations. Using a twostep procedure, the method first elicits non-incentivized reports of subjects' Personal Normative Beliefs regarding the most appropriate action from a set of possible options. In the second step, subjects are incentivized to predict the distribution of beliefs reported by others in the first step, thus capturing their normative expectations. However, the lack of incentives in the first step of the method introduces the potential for belief falsification. One possible motive for falsification is Social Desirability Bias. We explain how such bias could, in theory, influence measurement of norms under this method and report pre-registered experiments designed to induce biased disclosure of beliefs in the first step. Our experiments vary the threat of sanctioning by third- party monitors: in one treatment, respondents may wish to falsify their reported beliefs about the norm in a variant of the dictator game. Pre-registered results show a relatively small and non-significant effect of SDB. We explore the underlying conditions that make SDB more likely to threaten the identification of normative expectations. Exploratory results suggest an important role of awareness of the incentives to misreport in the first stage- the information asymmetry between respondents and third parties in our design. Researchers who plan to use this method to measure sensitive local norms should be aware of the conditions under which this potential bias is likely to materialize and design their studies to minimize it.