Measuring facets of worry: A LISREL analysis of the worry domains questionnaire

In the development of the Worry Domains Questionnaire (WDQ) for the measurement of nonpathological worry, (Tallis, Eysenck & Mathews, 1992. A questionnaire for the measurement of nonpathological worry. Personality and Individual Differences, 13, 161-168) Tallis er al. had used cluster analytical procedures to establish the number of worry domains. The resulting structure of the WDQ, however, was never adequately tested. This study therefore examined the WDQ's structure by use of confirmatory factor analysis comparing models of different factor structures. In the first sample of 466 participants, a five-factor model yielded the best fit to the data, characterized by highly correlated yet distinct domains of everyday worrying as they were originally proposed. This model was cross-validated with a second sample of 503 participants, showing stable factor loadings across samples. Whereas these analyses displayed a good fit of the five-factor representation for the item-based models, overall fit of all models was more prominent when items were aggregated (subscale models). Implications of the results and suggestions for future research are discussed. (C) 1997 Elsevier Science Ltd.