Inference from biased polls

People often attempt to present a positive image by overstating virtuous behaviors when responding to unincentivized “polls.” We examine whether others account for this “socially desirable responding” (SDR) when drawing inferences from such unincentivized responses. In an experiment, we incentivize “predictors” to guess others' choice behaviors across actions with varying social desirability. Predictors observe random subsamples of either (i) incentivized choices or (ii) hypothetical claims. The hypothetical claims exhibit systematic SDR and predictors are reasonably skeptical of them. However, their skepticism is not tailored to the direction or magnitude of SDR. This under-correction occurs even though subjects' stated sentiment toward the actions can predict SDR.