Using a small sample Q sort to identify item groups

A small sample of 40 second-year university students in Hong Kong were asked to perform a Q sort on the 36 items of a questionnaire used to assess six dimensions of student approaches to learning, the Learning Process Questionnaire. Participants were instructed to sort the items into as many groups as they liked with the only conditions being that all items within a group be as similar as possible in perceived meaning and that no item be placed into more than one group. By treating participants' responses as a form of multiple-choice data and by applying optimal scaling, three solutions were obtained. A two-dimensional plot of the optimal item scores for the first two solutions yielded 10 clearly defined item clusters suggesting a possible ten-factor model as opposed to the six-factor model found in studies of this questionnaire. The implications of using this approach to identify potential competing factor models, especially within a cross-cultural context, are discussed.