(PS2-54) The Reliability and Validity of the Social Cost Questionnaire: Assessing a Core Mechanism of Social Anxiety Disorder

Social anxiety disorder is maintained by the belief that negative social events have catastrophic consequences (Foa et al., 1996), and a reduction in social cost bias may be a key mechanism of CBT for social anxiety. The Social Cost Questionnaire (SCQ) is a 40-item instrument rating the cost of 20 negative social situations and 20 negative nonsocial situations. Although the SCQ is used in clinical studies to measure social cost as a mechanism of CBT, few studies have evaluated its psychometric properties. The goal of the current study was to evaluate the psychometric properties of the SCQ and its subscales of performance (social), non-performance (social), and nonsocial cost situations using data collected from participants with social anxiety (SOC), and without (healthy controls; HC).

This report is a preliminary analysis from an ongoing SOC treatment study that included 66 participants, ages 18-50 (Mage = 29.2 years, SD = 7.29). Participants completed a baseline assessment consisting of a semi-structured diagnostic interview (DIAMOND) and several self-report questionnaires, including the SCQ and measures of social anxiety severity (e.g., SPIN). In addition to scale reliability, concurrent validity was evaluated by (a) comparing SCQ subscale scores among SOC and HC, and (b) examining correlations between the SCQ subscale scores and measures of convergent (e.g., social anxiety severity assessed by the SPIN) and discriminant validity (e.g., overall disability). It was hypothesized that (a) only the SCQ social subscale scores (performance and non-performance) would be higher for participants with a SOC diagnosis compared to healthy controls, and (b) SCQ scores would be more strongly correlated with measures of social anxiety (e.g., SPIN and SAQ) than with measures of overall disability or quality of life.

Regression analyses indicated that individuals with a SOC diagnosis had significantly higher scores on the both the SCQ social (performance: B = 41.46,  t = 17.78, p< .001; non-performance: B = 36.02, t = 15.32, p< .001) and nonsocial scales (B = 37.09, t = 7.50, p< .001 ). SCQ social subscales (performance and non-performance) were more strongly correlated with measures of social anxiety severity (SPIN, r = .87, .92, p < .001; SAQ, r = .83, .92, p < . 001) than measures of disability or life events (WHODAS, r = .70, .70, p < . 001; QLES, r =- .722, -.723, p < . 001). Additionally, Cronbach’s alpha demonstrated high internal consistency for all three SCQ subscales (Cronbach’s α = .93 - .97).

In contrast with our hypothesis and inconsistent with previous findings, (Foa et al., 1996), there were statistically significant differences between SOC and HC on both the social and nonsocial events SCQ subscales. Nevertheless, the SCQ scales demonstrated strong scale reliability and concurrent validity. Based on these findings, the SCQ appears to have acceptable psychometric properties, though replication is needed in larger clinical samples. Recruitment is ongoing and analyses will be updated prior to conference presentation (expected n = 100).