The Misuse of Measurement in Economics

This paper is concerned with the problems that arise in economic analysis when ordinal variables are taken to be cardinal variables in models that require variables to be cardinally quantified for manipulative purposes. The main conclusion is that, in many important cases, this practice results in constructions that are without meaning, significance and power. Such a result applies, for example, to the efficiency wage model if effort, as many believe, is capable of only ordinal measurement. Similar problems arise (i) with the use of proxy variables, and (ii) even when prediction is the only goal.