A set of tests that might be reasonably used as provisional criterion measures in research on scientific thinking, particularly creative thinking, were developed and an assessment was made of the suitability of these tests as criterion variables from the standpoint of their psychometric properties. The Tests of Scientific Thinking are performance tests that simulate aspects of the job of a behavioral scientist. The tests are: Formulating Hypotheses, Evaluating Proposals, Solving Methodological Problems, and Measuring Constructs. The examinee proposes a number of solutions--not only the one considered best, but also others that should be considered. A scoring method was developed that requires the scorer to assign values to categories of responses rather than to make subjective evaluations. Six scores were studied. (1) average quality of all responses the examinee thinks are best; (2) average quality of all responses; (3) average quality of the best response by category scoring; (4) number of responses; (5) number of unusual responses; and (6) number of responses that are both unusual and of high quality. The tests were administered to about 4,000 graduate school applicants using an item sampling procedure. Test difficulty was found appropriate for advanced students and reliabilities were high enough to be useful. Factors analyses were performed to clarify the structure of the interrelationships among the various scores for the four tests. The tests seemed face valid, but evidence of construct validity is needed. (RC) (153pp.)
