Defection in the dark? A randomized‐response investigation of cooperativeness in social dilemma games

Abstract As previous research has demonstrated numerous times, humans show a robust tendency for cooperation. However, part—or indeed all—of this cooperativeness may be due to socially desirable responding. To address this problem, we propose and apply a new approach for the unbiased measurement of cooperativeness in social dilemma games. Specifically, we employ an extension of the randomized‐response technique (RRT). The RRT protects the privacy of respondents by adding random noise to their responses. It thus encourages more honest responding and thereby provides less biased estimates of sensitive attributes. In a large‐scale study with 2043 respondents we maximized anonymity in a one‐shot prisoner's dilemma game through use of the RRT. Comparing the prevalence rates for cooperation obtained via the RRT with those from direct self‐report showed that traditional direct questioning formats overestimate cooperation rates, with a relative bias of 18%. This finding suggests that to a considerable extent, self‐reported cooperation is due to socially desirable responding, rather than actual cooperativeness. However, our results also demonstrate that cooperation remains substantial even under conditions of maximized anonymity. Copyright © 2011 John Wiley & Sons, Ltd.