“Risk and punishment revisited — Errors in variables and in the lab”
To correct for measurement errors, one must first assess the magnitude of the error. To do this, one needs several measurements. Since repeated measurements are deemed expensive and complicated, researchers tend to ignore the problem.
Here we show that (a) the problem exists and (b) if risk attitudes are elicited with the help of a multiple price list task, then a simple solution exists. Seemingly inconsistent answers in the task can be exploited to assess and to correct for the measurement error.
We illustrate the problem with an experiment which studies the relation between risk aversion and punishment behaviour.
Keywords: Measurement error, risk, lab experiment, public good, Bayesian inference.
JEL: C91, D43, L41