Ideal point modeling of non-cognitive constructs: Review and recommendations for research

Abstract

Most psychological measurement models assume a monotonically increasing relation between the latent trait and the probability of responding. These models have historically been based on the measurement of abilities (e.g., cognitive), which have dominance properties. However, they are less appropriate for the measurement of non-cognitive constructs, or self-reported typical behaviors (e.g., attitudes, emotions, interests, personality), which have historically been assumed to have ideal point properties, or a nonmonotonic relation between the latent trait and the probability of responding. In this paper, we review the literature on ideal point modeling of non-cognitive constructs to present a theoretical framework that can help guide researchers on pertinent factors that may influence ideal point responding when assessing non-cognitive constructs. We also review the practical implications of not using ideal point response models for non-cognitive constructs and propose areas for research in non-cognitive construct assessment.