Under some circumstances, allowing examinees to choose which items to respond to may increase test validity. In this study, we explored how choice affected examinee performance and related characteristics for a test of the ability to generate hypotheses about a given situation. Four forms of this experimental test were randomly spiraled within a large sample of students taking the computer-based GRE General Test, with two of the forms allowing examinee choice. Results suggested that items were, in fact, differentially attractive to gender, ethnic, undergraduate major, and ability groups, indicating that, in some form, choice was operating. Further, performance was significantly higher when a given item was chosen than when it was required. Effects of choice on subgroup performance were generally inconclusive. Finally, there was some indication that internal consistency was improved by the choice option.