Reported is a study designed to investigate the convergent validity of the cognitive preference construct and to develop new techniques for analyzing cognitive preference data. The cognitive preference construct was invented to help evaluate the science curricula of the sixties based on the assumption that a student's intellectual approach to information was more important than ability to identify correct information. Four categories of intellectual approach as recall (R), principles (P), questioning (Q), and applications (A) are presented. Two new instruments to measure cognitive preferences were constructed and administered and a conventional instrument (the Science Cognitive Preference Inventory) was given to 71 elementary education majors. The results were analyzed using a multitrait-multimethod design. Developed were two statistical indices of distinctness of cognitive preference, one for ranking data and one for normative data. Correlations of Q and R with R, P, Q, and A scores across and within instruments, fulfilled most of the conditions for convergent validity. Correlations of P and A with R, P, Q, and A scores only partially fulfilled these conditions indicating weaknesses in the P and A modes. Correlations of Q and R strongly supported the interpretation of Q and R as opposite poles of one dimension which could be called "critical thinking." (Author/GS)