Individual differences in thinking-aloud performance

A key method in current research on thinking is to have subjects think aloud as they work on some task. Considerable individual differences usually emerge in the quantity and quality of thinking-aloud records. This article reports an experimental sudy of such individual differences. Subjects took Cattell’s 16-PF personality test, a self-relevation scale, and a test of verbal fluency. The 35 subjects then attempted three 8-puzzle tasks under thinking-aloud instructions. The resulting thinking-aloud records (protocols) were scored in terms of length, number of words per puzzle move, number of goal references, and number of goal references per move. The different protocol measures intercorrelated highly in each problem. Furthermore, each of the protocol measures correlated over different puzzles; that is, someone who produced a long protocol on one problem tended to produce a long protocol on other problems. However, none of the protocol measures correlated with any of the personality scores or with the verbal fluency measure. It is concluded that there is consistency within individuals in protocol quantity and completeness, and that this tendency merits further investigation.