Payoff Effects in Information Cascade Experiments.

Error rates are estimated using data from "information cascade" experiments. The econometric estimation assumes a logistic error structure and error rates are compared across three experimental treatments that differ only with respect to payoff structure. In a "no payoff" treatment subjects receive a fixed payment for participating in the experiment and earnings do not vary with decisions. In "payoff" and "double payoff" treatments earnings depend on each subject's decisions. The results indicate that rewarding correct decisions reduces the amount of decision error. However, increasing the payment for a correct decision does not reduce errors over the range of payoffs considered. Copyright 2001 by Oxford University Press.