Emotional Intelligence in Social Phobia and Other Anxiety Disorders

This is consistent with community sample psychometric data reported in the EQ-i:S manual, wherein alpha coefficients for the Adaptability subscale are the lowest for each age group for men, and tie for lowest for most age-groups in women.

Levene’s test showed homogeneity of variances across the four groups for the EQ-i:S total score and all subscales.

To rule out the possibility that differences between the SP group and the other two clinical groups’ Intrapersonal and Interpersonal EQ-I:S scores were due to state negative affect at time of self-report, analyses were conducted with data from a majority subset (74%, n = 218) of clinical participants that had completed the short Depression Anxiety and Stress Scale (DASS-21; Lovibond and Lovibond 1995) at the same time as the EQ-I:S. The DASS21 is a measure of current (last week) state affect, and its Depression and Anxiety subscales display sound psychometric properties in clinical anxiety samples (Antony et al. 1998a). We compared the three clinical groups’ scores on the two EQ-I:S subscales, with DASS Anxiety and Depression subscale scores entered simultaneously as covariates. With the state affect variables controlled for, differences between the groups remained significant for both Interpersonal, F (2, 213) = 6.20, p < .01 [DASS Depression β = −.41, t (213) = −4.94, p < .0001, Anxiety β = .10, t (213) = 1.22, n.s.] and Intrapersonal EQ-I:S scores F (2, 213) = 9.43, p < .0001 [DASS Depression β = −.42, t (213) = −5.46, p < .0001, Anxiety β = −.06, t (213) = −.82, n.s.].

Current (past month) GAF rating is conventionally used at our site, but for the present study’s purposes it was considered less meaningful than a rating of the individual’s highest level of functioning in the past year—an indicator of impairment conceivably less influenced by week-to-week fluctuations in symptoms and life events. However, past year GAF ratings were not assigned for the full duration of the data collection, thus sample sizes were reduced for analyses with the GAF (SP group n = 121; other anxiety group n = 88). As there was no difference between the PD (n = 39) and OCD (n = 49) groups in past-year GAF rating [t (86) = .33, p = n.s.], and there were no research questions about differences among SP and these specific diagnoses, their merger into a composite “other anxiety disorders” group seemed sensible for this and subsequent analyses. The control group was not included in these analyses as they were not administered the SCID-IV.

Tests of parallelism showed homogeneity of regression to be achieved across diagnostic groups for intrapersonal and Interpersonal subscale scores for all outcome variables analyzed with ANCOVA.

Current and lifetime comorbidity counts are ordinal data, and the distributions for both were positively skewed for both groups. Given the large sample sizes and the fact that the skew and spread was comparable across groups, both ANOVA and ANCOVA can be expected to be robust to violations of the assumptions of normality and homogeneity of variance (Tabachnick and Fidell 1996). As a further check, Levene’s tests confirmed homogeneity of variance across groups for both comorbidity outcomes, and nonparametric Kruskal-Wallis ANOVAs confirmed the ANOVA results.