Structural validity of the Insomnia Severity Index: A systematic review and meta-analysis

Abstract

The factor structure of the Insomnia Severity Index (ISI) is discussed, with examples being drawn from about 13 disparate reported models. A systematic appraisal of procedural details of studies investigating the dimensionality of the ISI was performed. The databases CINAHL, Cochrane Library, Embase, Medline, and PsycINFO were searched thoroughly, and all articles in English, published, and/or available online as of 14th March 2019, were included. Meta-analyses of Cronbach's alpha and structural validity measures were performed. Several inadequacies or omissions were found in analyses using the three groups of measures. Fifteen out of 20 studies did not employ either EFA or CFA, 10 studies did not report CFA, while five did not report EFA. The most common omissions related to multivariate normality, final model selection based on consideration of fit indices, parsimony, and failure to consider the theoretical construct in the final model selection. Meta-analysis showed that CFA outcome is more robust compared to EFA, two-factor solution is the vigorous presentation of dimensionality compared to a three-factor solution, and ISI has a high pooled Cronbach's alpha. Further validation of the disparate models of the ISI is needed with more comprehensive reporting guidelines, such as the one suggested in the present review.