Analyzing Complete Approval Voting Data under Thurstonian Framework

The method of approval voting is a commonly used voting procedure. In an approval voting task, each judge is usually asked to select a subset of the alternatives. For example, it is a common practice that the recruiting committee, after interviewing a set of job applicants, may apply approval voting to determine the qualification of the candidates at the first run of the hiring process. In analyzing social choices such as elections, the ”none” and ”all” responses are often discarded since they do not affect on the overall decision or conclusion. However, in understanding and modeling the approval voting mechanism or heuristics, it becomes preferable to include these two responses in the analysis as well. In this paper, we extend the current random utility approach to allow for the modeling of ”none” and ”all” responses and therefore provide a more complete account of the approval voting mechanism in the underlying choice process as well as the individual difference in their judgment. Empirical examples are given to illustrate the usefulness of our approval voting model formulation.