COMBINING MULTIPLE-CHOICE AND CONSTRUCTED-RESPONSE TEST-SCORES - AN ECONOMISTS VIEW

There exists considerable evidence from educational measurement theori sts that constructed-response questions add little new information abo ut student achievement beyond that provided by multiple-choice questio ns. The purpose of this study was to view this issue through the eyes of an economist, focusing on the bottom line of how many misclassifica tions of students would occur by replacing the constructed-response po rtion of the Advanced Placement exams in economics with more multiple- choice questions. A novel computer simulation methodology was employed , and it was discovered that some students do better on one type of te st than on the other and that this gives rise to a small but statistic ally significant number of misclassifications.