اختلالات عملکرد اجتماعی در اختلال اضطراب اجتماعی: واقعیت در طول مکالمه و درک مغرضانه در سخنرانی

Abstract Cognitive models emphasize that patients with social anxiety disorder (SAD) are mainly characterized by biased perception of their social performance. In addition, there is a growing body of evidence showing that SAD patients suffer from actual deficits in social interaction. To unravel what characterizes SAD patients the most, underestimation of social performance (defined as the discrepancy between self-perceived and observer-perceived social performance), or actual (observer-perceived) social performance, 48 patients with SAD and 27 normal control participants were observed during a speech and conversation. Consistent with the cognitive model of SAD, patients with SAD underestimated their social performance relative to control participants during the two interactions, but primarily during the speech. Actual social performance deficits were clearly apparent in the conversation but not in the speech. In conclusion, interactions that pull for more interpersonal skills, like a conversation, elicit more actual social performance deficits whereas, situations with a performance character, like a speech, bring about more cognitive distortions in patients with SAD.

Introduction Patients with social anxiety disorder (SAD) are concerned about flaws in their social performance, for instance, an anxious appearance (e.g., appearing nervous) and awkward social behavior (e.g., not knowing what to say). Cognitive models (Clark, 2001; Clark & Wells, 1995; Rapee & Heimberg, 1997) predict that patients with SAD overestimate their anxious appearance and social mishaps and underestimate quality of their social behavior. Indeed, research showed that socially anxious individuals suffer from biased perception of their social performance (Rapee & Lim, 1992; Stopa & Clark, 1993). That is, observers evaluated socially anxious individuals’ social performance as more positive than the socially anxious participants evaluated themselves. Yet, there might still be a core of truth in SAD patients’ concern about their social performance. In fact, a body of studies shows patients with SAD to perform worse in social tasks compared to control participants. This is found in both analogue (e.g., Beidel, Turner, & Dancu, 1985; Bögels, Rijsemus, & De Jong, 2002; Daly, Twentyman, & McFall, 1978; Lewin, McNeil, & Lipson, 1996; Thompson & Rapee, 2002; Twentyman & McFall, 1975) and patient samples (Baker & Edelmann, 2002; Fydrich, Chambless, Perry, Buergener, & Beazley, 1998; Stopa & Clark, 1993). Even though studies in this area are accumulating, it is still unresolved whether patients with SAD are mainly characterized with: (1) biased perception of their social performance, or by (2) actual performance deficits1. It is noteworthy that for most studies that evidenced social performance deficits in socially anxious participants, it remained unclear whether these deficits were due to social behavior, anxious appearance, or both. That is, many “social performance” rating scales possess items concerning visible anxious appearance, such as blushing, sweating, trembling, or appearing nervous. It could be argued that showing such physiologic responses is a deficit in social behavior. On the other hand, physiologic reactions are automatic responses that are not under voluntary control, whereas social behaviors such as smiling, nodding, asking questions are. In line, Bögels et al. (2002) found that anxious appearance and social behavior were two separate factors in observations of individuals that perform a social conversation. Accordingly, some studies found different outcomes for each of the two factors. That is, both Clark and Arkowitz (1975) and Bögels et al. (2002) found that high and low socially anxious individuals could be discriminated by an anxious appearance but not social behavior whereas Halford et al. (1982) found exactly the opposite pattern. In the last decades both conversation-like tasks (e.g., role plays, job interviews or getting acquainted) as well as speech tasks were subject to investigation in the social anxiety literature. When investigating biased perception and actual performance deficits in social tasks, it might be of value to discriminate between a conversation and a speech task. To illustrate, Rapee and Lim (1992) found no difference in actual social performance of SAD patients and controls during a speech, whereas other studies, which measured social performance (also) during conversation-like tasks, did find actual performance deficits in SAD patients (Alden & Wallace, 1995; Baker & Edelmann, 2002; Stopa & Clark, 1993). It seems plausible that a conversation calls upon more complex interpersonal social behaviors than a speech. For instance, during a conversation one needs to listen, ask questions, and respond to what others say, whereas during a speech one does not require interaction with the audience to such extent. Therefore, patients with SAD might show particularly deficits in social tasks that require these more difficult interpersonal social behaviors. Prior studies assessed biased perception of own social performance by calculating the discrepancy between observer ratings and ratings of participants’ own experience. However, instead of using participants’ rating of their own experience during the social task it would be more valuable to have participants estimate how they expect to be judged by their observers. That is, it is assumed that distorted self-perception is not specific for SAD but, for instance, also characterizes depression. Instead, SAD patients would specifically suffer from distorted perceptions of how others view them. The study of Strauman (1989) illustrates this nicely. Social anxiety was characterized by a discrepancy between how patients perceive themselves and what they believe others want them to be. In contrast, depression was marked by a discrepancy between how patients perceive themselves and how they ideally want to be. In other words, not living up to perceived standards of other people is related to social anxiety. Therefore, a more precise assessment of cognitive discrepancies in SAD appears participants’ ratings of how they believe observers will evaluate them, relative to observers’ actual evaluations. For the study reported here, we were interested whether patients with SAD were mostly characterized by biased perception or by actual performance deficits compared to control participants. We aimed to investigate these variables in both a speech and a conversation task and for both anxious appearance and social behavior. Therefore, we recruited patients with SAD (n = 48), and control participants (n = 27). They were asked to give an impromptu speech in front of two confederates (male and female) and to get acquainted with two other confederates (male and female). In order to assess biased perception regarding ones social performance, we let confederates rate participants’ social performance and simultaneously, let participants estimate these ratings of the confederates. Anxious appearance and social behavior were assessed separately for both biased and actual deficits.

. Results 3.1. Reliability of the SBA Consistency of the predicted ratings by the participants and the ratings by the confederates and video-observers of SBA-rating scale was high (range Cronbach α for anxious appearance: 0.90–0.92; social behavior: 0.91–0.92). The inter-rater reliability (ICC) for the two confederates on the SBA-rating scale was good for anxious appearance (speech: 0.76; conversation: 0.72) and for social behavior (speech: 0.81; conversation: 0.78). Moreover, each of the video-raters showed a moderate to good inter-rater agreement with the two confederates for anxious appearance and social behavior during speech and conversation (ICC, mean 0.84; range 0.59–0.93). The inter-rater reliability of the two video-raters was, as well, moderate to good (ICC mean 0.79; range 0.63–0.90). These inter-rater reliabilities compare well with those of similar previous studies ( Alden & Wallace, 1995; Bögels et al., 2002; Thompson & Rapee, 2002). In general we can conclude that the anxious appearance and social behavior ratings of the confederates were reliable. 3.2. Integrity of the confederates To assess integrity of the confederates the video-observer rated the behavior confederates toward the participants (see confederates in method section). Independent t-tests revealed that the confederates acted in the same way toward the SAD and the non-clinical control group. (speech: t(38) = 0.80, ns, conversation: t(38) = −0.72, ns). Thus, it can be concluded that integrity of the confederates was good. 3.3. Participants’ predicted ratings It was examined whether the patients with SAD would expect a worse social performance evaluation by their confederates than the non-clinical controls, which is denoted as “predicted rating.” The one-way ANOVA showed that the patients with SAD predicted to be rated worse by their confederates than the non-clinical controls on all measures: anxious appearance and social behavior during speech, F (1,74) = 15.23, p < 0.001; F (1,74) = 4.52, p < 0.05 and conversation F (1,74) = 12.01, p < 0.005; F (1,74) = 14.88, p < 0.001. Effect sizes of difference were large for both anxious appearance and social behavior during the conversation, large for anxious appearance during the speech, and medium for social behavior during the speech. In Table 1 means, standard deviations, and effect sizes of difference between the two groups are depicted. Table 1. Predicted, observed and discrepancy ratings of anxious appearance, social behavior and general social performance during speech and conversation for the patients with social anxiety disorder patients and the non-clinical control group Speech Conversation Patients with social anxiety disorder Non-clinical control group ES* Patients with social anxiety disorder Non-clinical control group ES* Anxious appearance Predicted 4.39 (1.37)a 3.12 (1.31)b 0.95 4.09 (1.34)a 2.88 (1.63)b 0.80 Confederates 3.58 (1.21) 3.18 (0.95) 0.37 3.47 (1.02)a 2.96 (0.85)b 0.54 Discrepancy+ 0.81 (1.62)a −0.06 (0.99)b 0.65 0.62 (1.47) −0.08 (1.57) 0.46 Social behavior Predicted 5.35 (1.51)a 6.06 (1.15)b 0.53 5.82 (1.13)a 6.82 (0.97)b 0.95 Confederates 6.70 (1.42) 6.77 (1.05) 0.06 6.51 (1.18)a 7.19 (1.10)b 0.60 Discrepancy+ 1.35 (1.44)a 0.70 (1.04)b 0.52 0.69 (1.23) 0.37 (1.18) 0.27 General social performance† Predicted 5.48 (1.35)a 6.45 (1.08)b 0.80 5.86 (1.08)a 6.95 (1.20)b 0.95 Confederates 6.57 (1.21) 6.79 (0.90) 0.21 6.52 (1.00)a 7.13 (0.92)b 0.63 Discrepancy+ 1.09 (1.40)a 0.34 (0.83)b 0.66 0.66 (1.17) 0.18 (1.24) 0.40 Note: Means with different superscript differ significantly (p < 0.05) from each other. * Effect sizes were calculated by Cohen (1987). d Statistic defined as (MSAD − Mnon-clinical controls)/S.D.pooled, where View the MathML sourceS.D.pooled=[(S.D.SAD2+S.D.non-clinical controls2)/2]. + Discrepancy = confederates rating minus predicted ratings of the participants. The higher the values the greater the underestimation of social performance by the participants. † For ‘social performance’ represents the mean of all items of the SBA-rating scale, the higher the values the better the social performance. Table options 3.4. Main analysis To examine the main research question whether patients with SAD are characterized primarily by biased estimations or by social performance deficits, a multivariate ANOVA was run with one between-subject factor (patients with SAD versus controls) and as dependent variables the eight variables described in the method section. These were, four variables concerning the observations of the confederates: (1) anxious appearance and (2) social behavior during the speech and (3) anxious appearance and (4) social behavior during the conversation, and four variables concerning the biased perception (discrepancy between actual and expected rating of confederates) of: (1) anxious appearance and (2) social behavior during the speech and (3) anxious appearance and (4) social behavior during the conversation. The MANOVA revealed the expected main effect for group F (8,66) = 3.25, p < 0.005. This main effect was unraveled by means of post hoc ANOVAs. Concerning the observations of the confederates, the patients with SAD were rated as showing a more anxious appearance and less adequate social behavior than the non-clinical controls during the conversation, F (1,75) = 4.76, p < 0.05; F (1,75) = 6.04, p < 0.05, but these differences were not significant during the speech, F (1,75) = 2.2, p = 0.15; F (1,75) = 0.04, p = 0.84. Concerning the discrepancy between actual and predicted evaluation, the patients with SAD showed a greater discrepancy than the non-clinical controls during the speech (anxious appearance: F (1,75) = 6.43, p < 0.05 and social behavior: F (1,75) = 4.21, p < 0.05). Also in the conversation the SAD patients showed greater discrepancy ratings than the non-clinical control participants. However, for social behavior this did not reach significance, F (1,75) = 1.3, p = 0.27, and for anxious appearance only a borderline effect was found, F (1,75) = 3.8, p = 0.06. In sum, patients with SAD were characterized by an underestimation of their performance mainly during a speech. During the conversation this underestimation for the SAD patients failed to reach significance. In addition to this biased perception of their social performance, patients with SAD showed actual social performance deficits during the conversation but not during the speech. The two types of social performance, anxious appearance and social behavior, showed generally the same pattern in differentiating the patients with SAD from the non-clinical controls. 3.5. Second main analysis: anxious appearance and social behavior combined It seemed that separating the two components of social performance, anxious appearance and social behavior, did not help to differentiate between the two groups. Therefore, we transformed social behavior and anxious appearance into a new variable “general social performance.” To construct this variable the mean across all items of the SBA was used for the predicted ratings by the participants and the ratings of the confederates during the speech and conversation: a higher rating indicating a better social performance. Four variables were calculated: ratings by the confederates during the (1) speech and (2) conversation, and discrepancy between confederates actual and predicted ratings of the participants during (3) speech and (4) conversation. Means and standard deviations are also depicted in Table 1. The MANOVA with these four dependent variables and group (patients with SAD versus controls) as between factor revealed again a main effect for group, F (4,70) = 5.06, p < 0.005. Post hoc ANOVAs indicated, in line with the previous findings, that during the conversation the confederates rated the SAD patients worse on social performance than the non-clinical controls, F (1,75) = 6.67, p < 0.05. Again, the confederates did not rate the two groups as significantly different during the speech, F (1,75) = 0.70, p = 0.41. In line with the previous analyses distinguishing anxious appearance and social behavior, patients with SAD underestimated their performance compared to non-clinical controls during the speech, F (1,75) = 9.83, p < 0.05, but for the conversation this difference was now borderline significant, F (1,75) = 4.05, p = 0.097. In sum, when combining anxious appearance and social behavior into one variable of general social performance, the conclusions mainly remained similar to those concerning the separated factors. That is, patients with SAD again showed actual social performance deficits during the conversation but not during the speech. In the same line, patients with SAD were again characterized by an underestimation of their performance during the speech. However, in contrast to the analyses separating anxious appearance from social behavior, during the conversation the underestimation of general social performance in the SAD group compared to the non-clinical control group started to approach significance.