Open AccessA STUDY OF THE PREDICTIVE VALIDITY OF THE TESTS OF SCIENTIFIC THINKING *
Author(s) -
Ward William C.,
Frederiksen Norman
Publication year - 1977
Publication title -
ets research bulletin series
Language(s) - English
Resource type - Journals
eISSN - 2333-8504
pISSN - 0424-6144
DOI - 10.1002/j.2333-8504.1977.tb01131.x
Subject(s) - psychology , test (biology) , variance (accounting) , ingenuity , consistency (knowledge bases) , predictive validity , applied psychology , social psychology , clinical psychology , statistics , mathematics , paleontology , geometry , neoclassical economics , accounting , economics , business , biology
The “Tests of Scientific Thinking” (TST) are free‐response job‐sample tests intended to measure aspects of scientific problem solving that require some degree of imaginativeness and ingenuity and that might therefore be used as dependent measures in research on “creativity.” The present study was designed to provide some preliminary evidence as to the validity of measures derived from these instruments. The TST and the GRE tests were compared with regard to their relationships to interests, self‐appraisals, and accomplishments of students during their first year of graduate work in psychology. These criterion variables were obtained from a questionnaire mailed to students near the end of the Spring, 1975, academic semester. Subjects were students who had taken the tests in October of 1973 for a study of the psychometric properties of the experimental tests. Difficulties in data analysis were created both by the item‐sampling character of the experimental test data, and by a relatively small rate of return of questionnaires from students who were actually attending graduate school in psychology. The estimated correlations for individual variables, therefore, reflect a substantial amount of error and can be regarded only as suggestive. It was possible to show, however, that the correlations do exhibit appreciable nonchance variance with regard to consistency across tests and across related questionnaire variables. The means, standard deviations, and reliabilities of the TST scores were in general about the same as those obtained in the psychometric study, as were correlations with GRE scores. The GRE tests were found to be more effective than the experimental instruments in predicting quality of the department attended, but the experimental tests were more effective in two other domains: (1) self‐appraisals of knowledge of psychology and skills in psychological activities, and (2) professional accomplishments such as research, publication, teaching, and tutoring. Of the several scores obtained from the new tests, those for number of Unusual responses and for number of Unusual‐High quality responses were the best predictors. The evidence is judged to justify further investigation of the validity of the experimental tests by other methods.