Bart Wilson has a recent working paper on deception in experimental economics. Like pretty much everything that I have read from Wilson, it’s primary purpose is to make the reader think. It is also completely non-technical, so I will not split this post into technical and non-technical sections.

Economists, unlike psychologists, do not deceive subjects in experiments. As Wilson outlines, the reason for this is  to make sure that, in future experiments, subjects will not be second guessing what might happen. If subjects were to be second guessing the experimenter, then they may not be responding to the monetary incentives that the experiment is designed to provide.

Wilson goes on to provide a series of examples, from the experimental literature, where the experimenters have gone close to deceiving the subjects (if you are at all interested in experimental design, it is worth reading Wilson’s paper. It is short.). In each case, Wilson finds that the experimenters did not engage in deception. In each case, I agree with Wilson’s judgement. Finally Wilson proposes the following test for deception “Did the experimenters mislead the participants by false appearance of action?” The implication that Wilson makes is that omission is not deception. It is hard to disagree.

Unfortunately, however, this is where Wilson stops.  Deception ,in and of itself, is not what we care about. What we care about is the preservation of the subject pool for future use. Asking “is X deception” is not enough. What we should be asking is “if subjects became aware of X either during or after this experiment, will it affect their behaviour in the next experiment?” Omission may very well fail this later test and should, therefore, be used with caution.