The Latent Factor Structure of Young's Early Maladaptive Schemas: Are Schemas Organized Into Domains?

Abstract

In previous research it has not been clarified whether the first-order schema factors of the Young Schema Questionnaire (YSQ) can be structured into higher order domains. We examined whether investigators' subjective choices between complex models of the YSQ or chance as opposed to clinical diversity of the samples may be responsible for the heterogeneity of results reported in the literature. We used confirmatory factor analysis to compare several a priori defined domain models in a sample of 542 undergraduate students (82.8% female; mean age 24.1 years) and 590 nonstudent adults (73.9% female; mean age 34.5 years) form an Internet survey. An additional Monte Carlo simulation study was performed to gain further insights on model selection. The analyses did not provide unequivocal support for the presence of a second-order domain structure. However, study findings suggested that the structure of the YSQ could be represented by a bifactor model including a first-order generic factor on which all items load and correlated first-order specific schema factors on which only the items load that were meant to measure the respective schemas. In the YSQ either several second-order domain structures are present that cannot be ranked by statistical measures alone or a first-order generic factor is present making second-order domains dispensable. Future research should include theoretical arguments and incorporate the clinical experience of practitioners.