Key points are not available for this paper at this time.
Abstract Criticism within the educational community of the multiple-choice format is in no short supply and seems to revolve around an alleged inability of multiplechoice tests to assess higher order thinking skills as well as constructed-response tests measure these skills. To investigate the credibility of such assertions, I constructed the exams for two measurement classes as half multiple choice and half constructed response; the exams contained equal numbers of items in each format written for the knowledge, comprehension, application, and analysis levels of Bloom's taxonomy. A pattern of generally high disattenuated correlations between multiple-choice and constructed-response measures within each taxonomic level was found, indicating that the two formats measure similar constructs at different levels of complexity. Factor analysis corroborated these interpretations.
Gregory R. Hancock (Sat,) studied this question.