A Revealed Preference Approach to the Elicitation of Political Attitudes: Experimental Evidence on Anti-Americanism in Pakistan

We develop a revealed preference method of eliciting political attitudes that overcomes many concerns associated with subjective responses to direct survey questions. We implement this method in an experiment in Pakistan that aims to understand the expression of anti-American attitudes. Following a standard personality survey, we offer subjects a bonus payment for the survey’s completion. Around one quarter of subjects are willing to forgo a 100 Rupee payment (roughly equal to one-third of a day’s wage) in order to avoid anonymously checking a box indicating gratitude towards the United States government for providing funds. Unbeknownst to subjects, we use their rejection of payment as an indicator of their expression of anti-American ideology; this measure, which we validate in the paper, mitigates concerns about experimenter demand effects or social stigma distorting subjects’ responses. Indeed, we find that the social context in which ideology is expressed matters: when individuals believe their choice to accept or reject the 100 Rupee payment will be observable by other subjects, they reject the payment significantly less often than in the private setting. The willingness to forgo payment is responsive to the payment size as well: significantly fewer subjects are willing to give up a 500 Rupee payment than a 100 Rupee payment. Using a discrete choice framework, we estimate an individual’s cost of publicly rejecting payment, in monetary terms, to be nearly 175 Rupees. We also estimate the fraction of individuals in our sample who would privately express anti-American views at zero monetary cost to be around 31%, quite close to what we find using a list experiment and anonymous direct survey questions. These findings both provide evidence on the determinants of the expression of anti-American views, and more generally provide a novel method for eliciting attitudes, which can be useful in other settings.