Estimating Rationality in Economics: A History of Statistical Methods in Experimental Economics

Experimental economists increasingly apply econometric techniques to interpret their data, as suggested the emergence of "experimetrics" in the 2000's (Camerer, 2003; Houser, 2008; Moffatt, 2015). Yet statistics remains a minor topic in experimental economics' (EE) methodology. This article aims to study the historical roots of this present paradox. To do so, we analyze the use of statistical tools in EE from early economics experiments of the 1940's-1950's to the present days. Our narrative is based on qualitative analysis of published papers for the earliest periods and on bibliometric and quantitative approaches for the more recent time period. Our results reveal a significant change in EE' statistical methods, from purely descriptive methods to more sophisticated and standardized techniques. Statistics now plays a decisive role in the way EE estimates rationality, particularly in structural modeling approaches, but it is still considered as a non-methodological, because purely technical, matter. Our historical analysis shows that this technical conception was the result of a long-run evolution of research tactics in EE, that notably allowed experimental economists to escape from psychologist's more re exive culture toward statistics.