Error prone inference from response time : the case of intuitive generosity

Response time is increasingly used to shed light on the process by which individuals make decisions. As mistakes may be correlated with response time it could, however, be misleading to use this measure to draw inference on preferences. To demonstrate we build on a recent literature, which uses response time to determine whether individuals intuitively are generous or selfish. Examining public good games researchers have shown that fast decision makers appear more generous than slow decision makers and this has been interpreted as evidence that generosity is intuitive and impulsive while selfishness is a calculated response (Rand et al. 2012; Nielsen, et al. 2014). Modifying the public good game to have an interior dominant strategy equilibrium we ask if the negative correlation between response time and giving is sensitive to the location of the equilibrium and whether it may result from mistakes. When the equilibrium is located below the midpoint of the strategy space we replicate earlier findings. However, when the equilibrium is located above the midpoint of the strategy space we get instead a positive correlation between response time and giving. While contribution distributions vary significantly by treatment for slow decision makers, these differences are not significant for fast decision makers. Fast decision makers are in both treatments more likely to make contributions that simultaneously lower individual and group earnings. We argue that the negative correlation between response time and giving rather than reflecting ‘spontaneous giving’, results from confused participants quickly selecting contributions that lie, on average, in the middle of the strategy space. Our results demonstrate that inference on preferences from response time requires that we take into account how mistakes are correlated with response time.