اختلافات میان عزت نفس ضمنی و صریح در نوجوانان مبتلا به اختلال اضطراب اجتماعی

Abstract Background and objectives Previous studies have found high implicit self-esteem (ISE) to prevail concurrently with low explicit self-esteem (ESE) in socially anxious adults. This suggests that self-esteem discrepancies are associated with social anxiety disorder (SAD). Given that the onset of SAD often occurs in adolescence, we investigated self-esteem discrepancies between ISE and ESE in adolescents suffering from SAD. Methods Two implicit measures (Affect Misattribution Procedure, Implicit Association Test) were used both before and after a social threat activation in 20 adolescents with SAD (14–20 years), and compared to 20 healthy adolescents who were matched for age and gender. The Rosenberg Self-Esteem Scale, the Social Cognitions Questionnaire and Beck Depression Inventory were administered as explicit measures. We expected discrepant self-esteem (high ISE, low ESE) in adolescents with SAD, in comparison to congruent self-esteem (positive ISE, positive ESE) in healthy controls, after social threat activation. Results Both the patient and control groups exhibited high positive ISE on both implicit measures, before as well as after social threat induction. Explicitly, patients suffering from SAD revealed lower levels of ESE, compared to the healthy adolescents. Conclusions This study is the first to examine ISE and ESE in a clinical sample of adolescent patients with SAD. Our results suggest that SAD is associated with a discrepancy between high ISE and low ESE, after a social-threat manipulation. The findings are discussed in relation to other studies using implicit measures in SAD and may provide a more comprehensive understanding of the role of self-esteem in adolescent SAD.

. Introduction According to cognitive models of social anxiety disorder (SAD, Clark & Wells, 1995; Rapee & Heimberg, 1997), dysfunctional self-views play an important role in the maintenance of the disorder among adults and adolescents (de Hullu, de Jong, Sportel, & Nauta, 2011). Specifically, negative self-views contribute to a discrepancy between perceptions of the self and of others' standards, ultimately leading to an increased fear of negative evaluation. Cognitive models of SAD assume that unconditional dysfunctional self-views are activated only when within social situations and are therefore considered unstable (Tanner, Stopa, & De Houwer, 2006). Due to their prominent role in maintaining SAD, previous studies have examined negative self-views, and specifically self-esteem among socially anxious adults (e.g., de Jong, 2002; Tanner et al., 2006). An important distinction is made in the literature between implicit (ISE) and explicit self-esteem (ESE). ESE refers to self-esteem of which the individual is consciously aware, whereas ISE is assumed to capture self-views of which the individual is unaware (Greenwald & Farnham, 2000), and are therefore not biased by self-representational tendencies (Paulhus, 2002). Thus, in order to obtain a comprehensive picture, it is important to examine both types of self-esteem among individuals with SAD (Buhrmester, Blanton, & Swann, 2011). Findings on implicit and explicit self-esteem in adult social anxiety seem to be partly at odds with the assumption of negative self-views in SAD, provided by cognitive theories (e.g., Clark & Wells, 1995; Roefs et al., 2011). Although not stated explicitly, from a cognitive perspective, self-esteem would be predicted to be negative in SAD. However, discrepant self-esteem (that is, simultaneously displaying high ISE and low ESE) has been reported in several studies on social anxiety. For instance, de Jong (2002) showed that high socially anxious women yield positive ISE in the Implicit Association Test (IAT, Greenwald, McGhee, & Schwartz, 1998), in contrast to negative ESE. Tanner et al. (2006) included a social threat activation (being told to give a speech) in order to investigate ISE in high and low socially anxious adults with the IAT, while controlling for depressive symptoms. The social threat activation was intended to activate negative self-views that are typical for social anxiety. Again, Tanner et al. found positive ISE in high socially anxious participants, which was less positive than that found in low socially anxious controls. Differences were found in ESE, which was significantly lower among high socially anxious individuals. A possible explanation for the diverging findings of high ISE and low ESE in social anxiety pertains to current dual-process models of self-esteem (e.g., Bosson, Brown, Zeigler-Hill, & Swann, 2003; Strack & Deutsch, 2004). According to these models, ISE and ESE reflect distinct, but associated constructs. ESE refers to beliefs or reflective processes, whereas ISE is related to impulsive affect-driven associative processes (e.g., Strack & Deutsch, 2004). Thus, these distinct processes may lead to discrepancies between ISE and ESE. The two studies on self-esteem in adult social anxiety, as well as on self-esteem in other disorders (Cockerham, Stopa, Bell, & Gregg, 2009; De Raedt, Schacht, Franck, & De Houwer, 2006; Franck, De Raedt, & De Houwer, 2008), support the relationship between self-esteem discrepancies and psychopathology. For instance, an investigation by Vater, Schröder-Abé, Schütz, Lammers, and Roepke (2010) revealed that self-esteem discrepancies are related to symptom severity in borderline personality disorder. According to Schröder-Abé, Rudolph, and Schütz (2007), and Schröder-Abé, Rudolph, Wiesner, and Schütz (2007), the combination of simultaneously exhibiting high ISE and low ESE is termed damaged self-esteem. Empirical evidence on discrepant or damaged self-esteem in adolescent social anxiety is sparse. This is surprising, since adolescence is a vulnerable period for the development of SAD (Wittchen & Fehm, 2003). Thus, examining the role of ISE and ESE within this age range should enhance our understanding of adolescent SAD. To our knowledge, only one analogue study has so far examined the role of negative self-esteem in socially anxious adolescents (de Jong, Sportel, de Hullu, & Nauta, 2011). In contrast to the self-esteem discrepancies found in the studies on adult social anxiety, this study reported a significant association between low ISE and social anxiety only for high socially anxious girls with very low levels of ESE. However, due to the nature of the sample, it remains unclear to what extent damaged self-esteem is related to adolescent SAD. Previous studies of self-esteem in social anxiety have used analogue samples rather than clinical ones (e.g., Tanner et al., 2006). In the present study however, we examined a sample of adolescents suffering from SAD. In addition, prior studies administered only a single implicit measure of self-esteem, the IAT (Roefs et al., 2011). By contrast, we used a novel implicit paradigm (Affect Misattribution Procedure, AMP, Payne, Cheng, Govorun, & Stewart, 2005) in addition to the IAT. The AMP is a promising implicit measure, which relies on intuitive affective judgments that correspond to everyday decision-making (Witthöft, Basfeld, Steinhoff, & Gerlach, 2011). This makes it particularly suitable for assessing automatic affect-driven ISE in SAD. An additional advantage is that it does not require working memory load, which has been shown to inhibit negative emotional processing (Van Dillen & Koole, 2007; Witthöft et al., 2011). In the present study, we used two implicit measures (AMP, IAT) and one explicit measure to examine self-esteem discrepancies after a social-threat-activation among adolescents both with SAD and without it, namely healthy adolescents. Since depression is a frequent comorbid condition in SAD among adolescents and young adults (Wittchen & Fehm, 2003), we controlled for depressive symptoms throughout all the analysis. Our central hypothesis was that adolescents with SAD would yield evidence of damaged self-esteem (high ISE, low ESE), compared to healthy controls, following the social threat activation. The latter would reveal a congruent self-esteem pattern (high ISE, high ESE). We also hypothesized that, after the activation of social threat, adolescents with SAD would display significantly lower ESE scores, in comparison to healthy controls. Finally, for adolescents with SAD, we expected the reactions on the AMP and IAT to correlate only weakly with explicit measures of self-esteem.

Results 3.1. Results for self-report measures and manipulation check The means and standard deviations of outcome variables, and the results from the MANOVA statistics for group comparisons, as well as effect sizes, are presented in Table 1. As expected, adolescents suffering from SAD differed significantly from healthy controls on all explicit variables. Adolescents with SAD displayed significantly lower explicit self-esteem, higher frequencies for socially phobic cognitions and also a stronger belief in these cognitions (all ps < .001). Moreover, the SAD group was significantly more depressed (p < .001). No differences were found regarding the strength of belief in the experimental manipulation between the two groups (SAD M = 91.3%, SD = 13.2 vs. healthy controls M = 84.3%, SD = 19.6; t(38) = 1.33, p = .19). In addition, no significant differences were revealed between the groups regarding educational level (χ2(1, 40) = .173, p = .68). Table 1. Symptom scores and self-esteem of socially anxious (SAD) and healthy (HA) adolescents. Variable Measure SAD (n = 20) HA (n = 20) F statistics from MANOVA (partial η2) M (SD) M (SD) Social anxiety SPAI 3.8 (.88) 1.2 (.70) 110.0*** (.74) Self-esteem RSES 23.6 (7.0) 35.4 (3.6) 43.3*** (.54) Frequency of negative social cognitions SCQ 2.9 (.87) 1.5 (.55) 34.7*** (.48) Strength of belief in negative social cognitions SCQ 45.7 (23.4) 10.1 (12.9) 35.4***(.48) Depression BDI-II 21.5 (15.7) 5.1 (5.1) 19.7*** (.35) Note. SPAI, Social Phobia and Anxiety Inventory; RSES, Rosenberg Self-esteem Scale; SCQ, Social Cognitions Questionnaire; BDI-II, Beck Depression Inventory. ***p < .001. Table options Since the groups differed with regard to depressive symptoms, the RSES and SCQ-scores were analysed by an additional MANCOVA, using the BDI score as a covariate. Depressive symptoms covaried significantly with the results for the RSES (F(1,36) = 72.59; p < .001; η2 = .67) as well as for the frequency (F(1,36) = 26.51; p < .001; η2 = .42) and strength of belief in negative social cognitions (F(1,36) = 27.57; p < .001; η2 = .43). However, the main effect of group, the direction of results and all single comparisons remained unchanged (Wilk's λ = .64, F(3, 34) = 6.51, p < .001, partial η2 = .37). Thus, after controlling for depressive symptoms, adolescents with SAD reported significantly lower levels of ESE, more negative social cognitions and believed more in them. 3.2. Results for the AMP The mean AMP ratings are displayed in Table 2. After exploring for possible violations of assumptions, a 4 (prime type: self-referent vs. positive vs. neutral vs. negative) × 2 (time: before vs. after the threat) × 2 (group: SAD vs. healthy controls) mixed MANCOVA with BDI as covariate, yielded a significant main effect of prime type on the AMP pleasantness rating (F(1.99, 73.78) = 4.96; p = .01; η2 = .12). No main effect of either “time (F(1, 37) < 1; p = .67)” or “group (F(1, 37) < 1; p = .79)” or any interaction effects (all ps < .05) were detected. Furthermore, the depression scores did not co-vary the AMP-ratings (F(1, 37) = 3.25; p = .08). Table 2. AMP and IAT scores in adolescents with social anxiety disorder (SAD) and healthy adolescents (HA) before and after a social threat activation (STA). Measure Primes SAD (n = 20) HA (n = 20) M (SD) M (SD) Before STA AMP Self-referent .63 (.22) .64 (.22) Positive .58 (.24) .68 (.25) Neutral .48 (.24) .58 (.26) Negative .46 (.24) .42 (.20) IAT .49 (.43) .59 (.24) After STA AMP Self-referent .62 (.26) .63 (.24) Positive .57 (.23) .70 (.25) Neutral .43 (.20) .48 (.25) Negative .51 (.26) .40 (.25) IAT .50 (.30) .53 (.26) Note. AMP = Affect Misattribution Procedure; IAT = Implicit Association Test; STA, social threat activation. Table options Bonferroni-adjusted post-hoc comparisons regarding the main effect of prime type suggested that target stimuli after positive primes (M = .63) were rated significantly more often as positive, in comparison to both negative (M = .45, p < .01) and neutral prime trials (M = .49, p < .05). This result was found before as well as after the social threat activation in both groups. The target ratings did not differ between self-referent prime trials and positive prime trials, indicating positive ISE (M = .63 vs. M = .63; p = 1.0) in both groups before and after the social threat induction. However, the target ratings after self-referent primes (M = .63) were found to be significantly more positive, compared to ratings following negative primes (M = .45, p < .01), but not statistically different from AMP-effects following neutral primes (M = .49, p = .08) in both samples, before and after a negative mood had been induced. 3.3. Results for the IAT Table 2 depicts the mean levels of ISE and standard deviations, as measured with the IAT in each group. A time (before vs. after the social threat) × group (SAD vs. healthy controls) mixed ANCOVA on the IAT-effects as the dependent variable, with the BDI as covariate, revealed no significant main effects of group (F(1,37) < 1; p = .78) or time (F(1,37) < 1; p = .91). Furthermore, the time × group interaction was not significant (F(1,37) < 1; p = .52). Depressive symptoms did not co-vary with the IAT-scores (F(1,37) < 1; p = .57). Thus, adolescents with SAD did not differ significantly from healthy controls in their IAT effects either before or after the social threat activation. 3.4. Association between implicit and explicit measures The AMP and IAT-scores were significantly positively interrelated after (r = .36, p = .01), but not before (r = .09, p = .30) the social threat activation. The AMP-scores were significantly positively interrelated (r = .70, p < .001) before and after the social threat activation, whereas the IAT-scores were not significantly associated (r = .27, p = .09) with each other. The results for all correlation analyses for adolescents with SAD are shown in Table 3. Table 3. Correlations between implicit and explicit measures in adolescents with SAD. Measure AMP after STA IAT after STA RSES SCQ-frequency SCQ-belief SPAI AMP after STA .45* .41 −.45* −.50* −.41* IAT after STA .45* .11 −.04 −.08 .04 RSES .41 .11 −.86*** −.83*** −.80*** SCQ-frequency −.45* −.04 −.86*** .96*** .86*** SCQ-belief −.50* −.08 −.83*** .96*** .83*** SPAI −.41* .04 −.80*** .86*** .83*** Note. AMP, Affect Misattribution Procedure; IAT, Implicit Association Test; STA, social threat activation; RSES, Rosenberg Self-esteem Scale; SCQ, Social Cognitions Questionnaire; SPAI, Social Phobia and Anxiety Inventory. *p < .05; ***p < .001. Table options To examine the relationship between ISE and ESE, two-tailed Pearson correlations were computed between the IAT after the social threat activation (AMP, respectively) and RSES within each group, since the RSES was administered only after the social threat induction. The AMP-score after self-referent primes and RSES, as well as the IAT and RSES, were not found to correlate significantly in SAD patients (see Table 3). Accordingly, no significant associations between ISE and ESE were found for healthy adolescents (RSES-IAT: r = −.02, p = .46; RSES-AMP: r = −.25, p = .14). In adolescents with SAD, the AMP-score yielded significant inverse relationships with the frequency of negative social cognitions (SCQ), the strength of belief with respect to these cognitions (SCQ) and social anxiety symptoms (SPAI, see Table 3; all p < .05). However, this did not apply to the healthy control group. For the IAT-score, no significant correlations could be found for either group. Regarding the relationship between ESE and socially phobic questionnaires (SCQ, SPAI), high positive and significant relationships were found (all p < .001, see Table 3).