The latent structure of social anxiety disorder and the performance only specifier: a taxometric analysis

Abstract

Social anxiety disorder (SAD) is often treated as a discrete diagnostic entity that represents a naturally occurring class, though empirical evidence largely supports a dimensional conceptualization of social fears. Further, the inclusion of a “performance only” specifier in the DSM-5 implies that individuals who experience intense social anxiety exclusively in performance situations are distinct from those with broader social fears. The purpose of the present research was to examine the latent structure of SAD and the DSM-5 “performance only” specifier in a large nonclinical sample (n = 2019). Three taxometric procedures (MAXCOV, MAMBAC, and L-Mode) were applied to indicators derived from two commonly used measures of social anxiety. Results yielded convergent evidence indicating that social anxiety exhibits a dimensional latent structure. Further, social performance anxiety demonstrates continuous relationships with milder social fears, suggesting that the “performance only” specifier may not represent a discrete entity. The implications of these findings for the assessment, diagnosis, classification, and treatment of social anxiety are discussed.

Keywords:

• Social anxiety
• performance
• anxiety
• dimensional
• taxometric

Notes

1. Taxometric analyses were also conducted using composite indicators representing the 3 factors originally identified by Safren et al. (1998). The results generally supported a dimensional solution, with each of the 4 MAXCOV and L-Mode plots exhibiting a characteristic dimensional shape and the mean CCFI score being .37 (i.e. less than .45), though the MAMBAC procedure did not pass the initial suitability test as the raters could not discriminate between simulated taxonic and dimensional plots. In addition, MAXEIG analyses, a multivariate extension of MAXCOV that can handle larger numbers of indicators, were conducted using all of the SPS and SIAS items excluded by our EFA as indicators. After discarding several items that failed to meet minimum validity criteria (>1.25; Meehl, 1995), results of the MAXEIG analyses provided strong support for a dimensional solution (CCFI = .11), suggesting that the findings of the primary analyses were not an artifact of the indicators derived by the EFA. However, to conserve space and ease interpretability of findings, only results generated using the indicators derived via EFA in the present study are presented in the text.

2. MAXEIG analyses were also conducted using each of the individual SPS items as indicators of “performance only” social anxiety. After excluding nearly half of the items due to failure to meet minimum validity criteria, results supported the dimensional results reported using the empirical approach to indicator selection (SPS CCFI = .40).