Dimensional Structure of the Symptom Checklist-90-R: A Network Analysis with a Combined Clinical and Community Sample

Notwithstanding its growing popularity and advantages over exploratory and confirmatory factor analysis, no studies have as yet employed psychometric network analysis with clinical subjects for attempting to resolve the dilemma about the dimensional structure of one of the most widely employed self-report measures of psychopathology: the Symptom Checklist-90-R [SCL-90-R]. This is the first study to do so. Clique Percolation, a network community detection algorithm that can accomodate overlapping nodes (symptoms), was applied to SCL-90-R self-ratings of a combined psychiatric patient and community sample ( n = 2,069). Thirteen replicable and reliable primary communities (dimensions) that also differentiated the patient from the community sample in the predicted direction were yielded: Dysphoria , Interpersonal Sensitivity , Negative Self-Evaluation , Cognitive-Performance Deficits , Anxiety-Fear , Agoraphobia , Hostility , Social Anxiety , Somatic Anxiety , Cardiopulmonary Symptoms , Mistrust , Thought-Related Delusions , and Existential Distress . Visual inspection of the conditional associations between these 13 primary communities suggested the possibility of three higher-order dimensions: Panic with agoraphobia, Depression, and Interpersonal difficulties. The primary dimensions of Hostility and Thought-Related Delusions were the two least central dimensions in this visual alignment. Thirteen primary symptom dimensions are now available for further research on measurement invariance across relevant subject parameters (e.g., gender, age, clinical status) and for obtaining data on symptom status at initial patient interview and on patient reported outcome (to monitor treatment progress). Tentatively, in the Hierarchical Taxonomy of Psychopathology [HiTOP], Hostility could be located in the Antagonistic Disinhibited Spectrum and Thought-Related Delusions in the Thought Disorder Spectrum. The remaining 11 primary dimensions could all be allocated to the HiTOP Internalizing Spectrum.