During the past decade, the experimental economics method increasingly has been used to study the impact of tax policy on taxpayer behavior. Experimental economics in taxation typically tests tax applications of expected utility and psychological theory by creating a real microeconomy in the laboratory. A key requirement is strict control over the parameters of the experimental setting and subject preferences. One generally accepted procedure to achieve experimental control is to avoid references to real‐world phenomena in instructions to subjects. The reasoning underlying this procedure is that if subjects associate the experiment with real‐world phenomena, they may make decisions based on values associated with the real‐world context instead of the rewards and penalties of the microeconomy. Despite this, several recent tax‐reporting experiments that otherwise conform to the experimental economics method have used explicit tax terminology. In discussing the results of these experiments, the authors and commentators stated that the results probably were not affected by the tax context.
We conducted an experiment to further examine whether the results of a tax‐reporting experiment were affected by context. We found that subjects reported significantly more income when the context of the experiment was tax than when the context was nontax. This effect differed, however, depending on the age of the subjects. While subjects 25 years of age and older reported approximately twice as much income in the tax context as they did in the nontax context, subjects under 25 reported only slightly more in the tax context than in the nontax context. These results provide evidence that role playing based on individual subject characteristics occurred when contextual cues were provided.