Person-centred psychometrics and cognitive style: moving from variables to a person-centred approach

Abstract

Cognitive style refers to the way in which we prefer to think and process information. Within the framework of individual differences, cognitive style has much to offer with regards to our understanding of how people think. The field of cognitive style has been plagued by poorly constructed instruments based on disparate and incomplete theory. One instrument, the Ways of Thinking, offers an alternative conceptualisation of cognitive style. Drawing on what has been learned from the study of personality, the Ways of Thinking instrument was developed through an application of the psycholexical approach. Early modelling of the Ways of Thinking showed much promise, while mixed method research added to the construct validity of the instrument, further suggesting that there was evidence for three integrative typologies – Realists, Ideaists, and Undifferentiated thinkers. This thesis aims to start where previous validation research into the Ways of Thinking instrument finished. Given that the three qualitative typologies were not reflected in the hierarchical statistical model of the instrument, a person-centred perspective to construct validation was considered. Building on a foundation of classical test theory and item response theory, variable-centred validation studies were completed; a short-form version of the instrument was created, and the temporal stability of both versions was established; before finally considering three person-centred methodologies. The goal was to explore higher order types of people by using the Ways of Thinking facets in conjunction with transposed principal components analysis, cluster analysis, and latent class analysis. The results of the variable-centred analyses suggested that the Ways of Thinking instrument generally demonstrated excellent measurement model fit against both classical test and item response theories. The short-form version explained an average of 83% of the variance in the long-form instrument, while temporal stability across three substantial time periods was acceptable. The results of the person-centred analyses were remarkably similar given the different mathematical models underpinning the methods, with clear evidence in support of a divide between Realist and Ideaist thinkers, and subtle differences in subtypes also emerged. The predictive utility of the Ways of Thinking instrument was explored using personality and occupational interests as outcomes. As expected, the Ways of Thinking instrument demonstrated strong relationships in theoretically consistent directions with a measure of the Big Five personality factors and a measure of Holland’s RIASEC model. The short-form version demonstrated very little loss in predictive utility compared to the long-form version. The person-centred typologies also proved to be of considerable predictive utility, which was an important aspect of construct validity to demonstrate. Implications, limitations, and future directions for research were explored.