Endogeneity, Incompleteness, and Choice Set Construction

• In Event: Quantitative Methods and College Access

Abstract

When applying rational choice theory to study college choice, a critical issue is to correctly model the sets of colleges available to students, or the choice sets. The construction of choice sets, however, is an unsettled issue. Some researchers argued that the empirically observed choice sets must be remediated before use because of endogeneity and incompleteness. Others contended that neither would bias the estimate when the independence of irrelevant alternatives assumption is met. This article contributes to the debate by borrowing insights from psychometrics. It demonstrates that the college choice problems can be reconceptualized as psychometric problems. It then offers intuitive explanations about why endogeneity and incompleteness do not bias estimates. Supporting evidence from a simulation study is also provided.

Author

• Yiran Chen, University of Michigan - Ann Arbor