J. Neil Bearden
Associate Professor of Decision Sciences
The authors study a class of sequential selection and assignment problems in which a decision maker (DM) must sequentially assign applicants to positions with the objective of minimizing expected cost. In modeling this class of problems, the authors assume that on each period the DM is only informed of the rank of the present applicant relative to the applicants that she previously observed and assigned. We first present the optimal decision policy that we subsequently use as a normative benchmark, and then report results from three experiments designed to study sequential assignment behavior.In comparing the aggregate results from all three experiments to the optimal decision policy, the authors identify a systematic bias, called the middleness bias, to over-assign applicants to intermediate positions. The results also reveal a strong bias for early applicants to be over-assigned to important positions.