Newsvendor pull-to-center reconsidered

In the newsvendor problem, a pull-to-center effect has been asserted, whereby subjects are said to order a quantity between the mean of the demand distribution and the expected profit-maximizing quantity. These claims have only been examined using group-level aggregate statistics. Looking at individual-level data from a previously published study and a new experiment, the current paper shows that while pull-to-center is present in aggregate data, it does not adequately describe the population of individual decision makers, who are found to be highly heterogeneous. Methodological implications and future research directions are discussed.