Many faces of bankers' identity: how (not) to study dishonesty

In a recent Nature article, Cohn et al. (2014, henceforth CFM) boldly claim that their results “suggest that the prevailing business culture in the banking industry weakens and undermines the honesty norm” (p. 1). The main empirical finding of the CFM paper is that a group of bank employees that answered a series of questions about their work subsequently reported higher proportion of wins in a coin flipping task (58.2%) than a control group of bank employees (51.6%) that answered questions of a similar form but focused on their leisure activities instead of banking. Because one can on average correctly guess the outcome of only half of the flips, a higher rate of success reported by a group indicates that some of its members cheated. From this result the authors infer that asking bankers about their work makes their professional identity more salient, which leads them to cheat more. That in turn is supposed to suggest that “prevailing business culture in the banking industry favors dishonest behavior” (CFM, p. 3). We do not want to question the reality of their empirical findings. However, in line with previous critiques of priming research (Stafford, 2014), we want to point out certain limitations of the inferences which can be drawn from the empirical results: First, the observed dishonest behavior after the banking identity prime does not necessarily mean that there are social norms encouraging dishonesty in the banking industry. Instead of norms, a negative stereotype of bankers’ dishonesty might be responsible for the observed effect: other stereotypes were shown to influence behavior—for example priming of the “hooligan” stereotype decreases performance in a trivia quiz (Dijksterhuis and van Knippenberg, 1998). Second, it is not even necessary that the observed difference in cheating had been caused by the priming of the professional identity. Either the other group could have been primed to cheat less, or other related concepts could have been primed in the experimental condition together with the banking identity and caused more cheating. Our goal is to highlight the fact that the study of cultural norms on behavior using indirect methods often precludes us from making strong conclusions, because alternative explanations are plentiful and hard to rule out. We offer suggestions how future studies can build on the CFM design and more reliably answer the question whether the social norm in banking really supports dishonesty.