پیش بینی کوتاه مدت عاطفه مثبت در افراد مبتلا به اختلال اضطراب اجتماعی: نقش صفات شخصیتی انتخاب شده و استراتژی تنظیم احساسات
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|38863||2015||10 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Anxiety Disorders, Volume 34, August 2015, Pages 53–62
Abstract Recently, research has provided support for a moderate, inverse relationship between social anxiety and dispositional positive affect. However, the dynamics of this relationship remain poorly understood. The present study evaluates whether certain personality traits and emotion regulation variables predict short-term positive affect for individuals with social anxiety disorder and healthy controls. Positive affect as measured by two self-report instruments was assessed before and after two tasks in which the participant conversed with either a friend or a romantic partner. Tests of models examining the hypothesized prospective predictors revealed that the paths did not differ significantly across diagnostic group and both groups showed the hypothesized patterns of endorsement for the emotion regulation variables. Further, a variable reflecting difficulty redirecting oneself when distressed prospectively predicted one measure of positive affect. Additional research is needed to explore further the role of emotion regulation strategies on positive emotions for individuals higher in social anxiety.
1. Introduction Although it was initially proposed that diminished positive affect was specific to depression and differentiated the Axis I mood and anxiety disorders of the Diagnostic and Statistical Manual of Mental Disorders, (DSM-IV-TR; American Psychiatric Association, 2000), a growing body of work has indicated that this affective pattern may also characterize individuals with social anxiety disorder (e.g., Brown et al., 1998 and Watson et al., 1988a). In a meta-analysis evaluating the magnitude, consistency, and specificity of the relationship between social anxiety symptoms and positive affect, Kashdan (2007) found a moderate, inverse relationship between the constructs (r = −.36; 95% CI: −.31 to −.40). Additionally, comorbidity with depressive symptoms or other emotional disturbances did not account for this result ( Kashdan, 2007). In addition to lower levels of dispositional positive affect, higher social anxiety is also associated with less frequent, lower intensity, and less lasting positive experiences, as well as fear responses to overtly positive social experiences and lower self-reported expression of positive emotions as compared to lower social anxiety ( Kashdan et al., 2011 and Turk et al., 2005). Collectively, the extant literature has provided support for a link between excessive social anxiety and reduced hedonic activity. In one such study, Eisner, Johnson, and Carver (2009) found that even after controlling for lifetime depressive symptoms, tendencies to endorse strategies aimed at diminishing or discouraging positive affective states (i.e., dampening) were positively related, whereas endorsement of strategies aimed at enhancing and sustaining positive affective states (i.e., savoring) was inversely related, to symptoms of social anxiety disorder. Eisner and colleagues’ findings provided initial evidence that people with social anxiety disorder are not only characterized by lower trait positive affect, but may also employ maladaptive emotion regulation strategies that are detrimental to maintaining this affect. Furthermore, there has been some research showing that individuals higher in social anxiety tend to use strategic thought suppression, a form of emotional response regulation that involves inhibiting ongoing expressive behavior (Gross, 1998) and has been linked to the dampening of positive emotions (Gross and John, 2003, Gross and Levenson, 1997 and Spokas et al., 2009). Werner, Goldin, Ball, Heimberg, and Gross (2011) reported that in addition to greater use of avoidance and expressive suppression, individuals with social anxiety disorder reported less success in implementing cognitive reappraisal and expressive suppression strategies during a laboratory speech task and two social anxiety-evoking situations. Moreover, it has been proposed that people with excessive social anxiety are characterized by deficits in understanding of, and negative reactivity to, emotions (e.g., Turk et al., 2005) Taken together, these studies have laid the groundwork for further investigation into how the relationship between social anxiety and positive experiences may vary as a function of how people manage their emotions in everyday life. To our knowledge, there have only been two studies directly examining short-term positive affect variation in individuals with excessive social anxiety. In the first, an experience-sampling assessment of positive emotions and events over a 21-day period, Kashdan and Steger (2006) found that individuals higher in trait social anxiety reported fewer positive events on days when they experienced greater social anxiety and used emotional suppression; however, this effect was not present for individuals lower in trait social anxiety. Further, participants generally reported the most intense positive emotions on days when they were both least socially anxious and most accepting of emotional experiences irrespective of dispositional social anxiety. Similarly, Kashdan and Breen (2008) found support for interactions between social anxiety and the emotion regulation tendencies of suppression, positive emotional expression, and negative emotional expression in predicting changes in positive emotions over 12 weeks. Whereas less suppression and greater expression (regardless of valence) predicted greater changes in positive emotions for individuals with lower social anxiety, none of these variables significantly predicted change in positive emotions for individuals with higher social anxiety. Notably, the presence of a relationship between suppression and positive emotions exclusively for those lower in social anxiety directly contradicts the findings of Kashdan and Steger (2006). However, the combination of relatively low social anxiety along with tendencies to be more accepting and expressive of emotions led to the highest levels of positive emotions, which is consistent with Kashdan and Steger (2006). Despite conflicting findings involving suppression, the results of this pair of studies with regard to emotional expression are consistent with research suggesting that greater tendencies to accept and openly express emotions are integral to the increased positive psychological functioning seen in individuals lower in social anxiety (e.g., Hayes et al., 2006 and Hayes et al., 1996). As an extension of this budding literature, initial work has supported a relationship between emotion regulation strategies and personality traits that, in turn, may have an interactive effect on both positive and negative affective experiences. Collectively, the available research has provided consistent support that neuroticism is associated with maladaptive emotion regulation strategies, and extraversion is associated with adaptive strategies (e.g., Davies et al., 1998 and Wood et al., 2003). Likewise, neuroticism correlates positively with increases in negative emotions in response to unpleasant stimuli, whereas extraversion correlates positively with increases in positive emotions in response to pleasant stimuli (Gross, Sutton, & Ketelaar, 1998). Furthermore, research has provided evidence of relationships between social anxiety disorder and these personality traits, with low neuroticism and high extraversion being significantly related to the disorder in epidemiological as well as nonclinical samples (Bienvenu et al., 2001, Bienvenu et al., 2004 and Trull and Sher, 1994). Finally, a recent investigation into the facets of extraversion/positive emotionality (E/PE; e.g., sociability) revealed that social anxiety was broadly related to all facets of E/PE after controlling for the higher-order internalizing factor and the overlap among the individual facets (Naragon-Gainey, Watson, & Markon, 2009; Naragon-Gainey and Watson, 2011). In sum, there is a well-established relationship between diminished positive affect and social anxiety disorder, but the dynamics of this relationship are not well understood. The present study offers an extension of previous work by evaluating whether certain emotion regulation and personality trait variables prospectively predict changes in state positive affect for individuals with social anxiety disorder as well as demographically equivalent control participants. Positive affect was assessed before and after two behavioral tasks in which the participant conversed with either a friend or a romantic partner. The content of the interactions was controlled to elicit emotional responses; the first task was aimed at promoting social support, whereas the second task was aimed at promoting conflict within the dyad. It is important to note that the current study focused specifically on interpersonal relationships that elicit emotions as opposed to the relationships themselves. For examinations of relationship quality in this sample, please see Rodebaugh et al. (2013), Rodebaugh et al. (2014). Our first hypothesis was that extraversion would prospectively predict positive affect after the support task, but not necessarily after the conflict task. This hypothesis reflects research showing that individuals are characterized by higher positive affect when acting more extraverted (Fleeson, Malanos, & Achille, 2002). We also proposed that whereas the social support task would lead to increased expression of extraversion to expand positive affect, the conflict task would involve a greater focus on retaining positive affect, indirectly resulting in fewer opportunities for the expression of extraversion. That is, given that the conflict task was aimed to be contentious, increased positive affect seemed unlikely. Our second hypothesis was that difficulties in emotion regulation and the personality trait of neuroticism would prospectively predict positive affect after the conflict task. Taken together, the first two hypotheses predict not only that the magnitudes of these predictive relationships will differ significantly across assessments, but also that the selected variables will be significant predictors of positive affect at specific time points as a consequence of the behavioral tasks. We did not generate hypotheses concerning positive affect fluctuations between the post-support and pre-conflict measurements, a break period that lacked a systematic intervention and therefore would likely be characterized by substantial error variance. Thirdly, we predicted that the diagnostic groups would show differential endorsement of the emotion regulation variables based on theoretical adaptiveness, although we did not expect the relationships between the predictors and positive affect to differ across diagnosis. Our fourth and final hypothesis was related to relationship type (i.e., friend or romantic partner). We theorized that the romantic relationship would magnify all effects, such that having an emotionally salient interaction with one's significant other would result in greater fluctuations in positive affect regardless of the nature of the conversation (i.e., social support or conflict). Accordingly, we hypothesized that there would be an interaction between relationship type and each of the emotion regulation and personality variables in predicting positive affect, regardless of whether the individual variables are significant prospective predictors of positive affect following the behavioral tasks.
نتیجه گیری انگلیسی
3. Results 3.1. Tests for differential endorsement across groups To test our hypotheses regarding differential endorsement of the emotion regulation variables between the GSAD and NOSAD groups, we conducted a chi-square difference test comparing a model in which the means of these variables were constrained (to be the same value) across group to a model in which the means were unconstrained. Consistent with prediction, the constraints resulted in significant decrement to fit (p < .001), indicating that the diagnostic groups differed substantively in their endorsement of the emotion regulation tendencies. All effects were in the expected direction, with the GSAD group showing higher scores on the nonacceptance of emotional responses and difficulties engaging in goal-directed behavior factors of the DERS compared to the NOSAD group, and the NOSAD group showing higher scores on the nonreactivity subscale of the FFMQ relative to the GSAD group. See Table 1 for descriptive statistics across diagnostic group. Table 1. Frequencies and descriptive statistics for social anxiety disorder (GSAD) and no social anxiety disorder (NOSAD) participants. GSAD group (n = 65) NOSAD group (n = 49) Age 40.72 (13.70) 36.86 (13.63) Women 47 (72.3%) 35 (71.4%) Race and ethnicity White 35 (53.8%) 29 (59.2%) Asian 2 (3.1%) 1 (2.0%) Black 22 (33.8%) 18 (36.7%) Multiracial 5 (7.7%) 1 (2.0%) American Indian 1 (1.5%) 0 (0%) Hispanic 1 (1.5%) 2 (4.1%) Difficulty in emotion regulation scale (DERS) Nonacceptance of emotional responses 14.84 (5.95) 9.23 (2.62) Difficulties engaging in goal-directed behavior 15.26 (4.92) 10.38 (4.38) The five factor mindfulness questionnaire (FFMQ) Nonreactivity to inner experience 20.10 (4.49) 23.98 (4.96) Mini-international personality item pool (Mini-IPIP) Extraversion 8.19 (3.03) 15.23 (3.01) Neuroticism 13.66 (3.33) 8.85 (3.08) Positive and negative affect schedule (PANAS) Trait assessment 28.30 (8.25) 36.73 (6.86) Pre-support (T1) 28.30 (9.15) 30.53 (9.65) Post-support (T2) 30.05 (9.95) 31.67 (10.07) Pre-conflict (T3) 29.32 (10.26) 30.71 (10.54) Post-conflict (T4) 30.05 (11.19) 32.45 (10.51) Self-assessment manikin (SAM) Pre-support (T1) 3.91 (1.71) 3.13 (1.41) Post-support (T2) 3.49 (1.90) 2.90 (1.65) Pre-conflict (T3) 3.77 (1.83) 3.14 (1.65) Post-conflict (T4) 3.86 (2.28) 3.18 (1.68) Note: Table 1 is reflective of the complete data and thus does not include values obtained through multiple imputation. Table options 3.2. Tests of initial autoregressive models The first set of multiple-group models incorporated three different regressions aimed at prospective prediction of positive affect at T2, T3, and T4. We regressed each individual time point on the previous time point (known as an autoregressive parameter) as well as the hypothesized emotion regulation and personality trait predictor variables. Additionally, to reflect the likelihood that positive affect at time points prior to a task would be related to one another and that positive affect at time points following a task would also be related to one another, T1 was allowed to correlate with T3 and T2 was allowed to correlate with T4. All predictor variables were also allowed to correlate with one another as well as T1.3 Tests of this original hypothesized model for the PANAS resulted in suboptimal fit indices, particularly the RMSEA (CFI = .96, TLI = .93, RMSEA = .13). After conducting follow-up tests aimed at addressing the inflated RMSEA, we found that that allowing T1 to correlate with T4 markedly improved model fit (CFI = .99, TLI = .98, RMSEA = .07). Parallel tests using the SAM data yielded a similar pattern of results (original: CFI = .95, TLI = .92, RMSEA = .07; revised: CFI = .96, TLI = .94, RMSEA = .07). Accordingly, subsequent analyses testing constraints across group and time point included this path.4 3.3. Tests constraining parameters-PANAS 3.3.1. Constraints across diagnostic group These analyses were conducted to test our hypothesis regarding a lack of differential predictive relationships between the emotion regulation and personality trait variables and positive affect across diagnostic group. We conducted chi-square difference tests comparing models in which all predictor paths were constrained across group to models in which these paths were unconstrained in order to examine whether the predictive relationships between the variables and state positive affect vary based on diagnosis. For the PANAS data, these tests resulted in nonsignificant p-values in all five of the imputed data sets (ps > .213). This set of findings can be interpreted to reflect that the constraints did not lead to significantly worse model fit, consistent with our hypothesis that the predictive relationships would not differ across group. Accordingly, we retained all constraints across group in the analyses detailed below, reflecting our data analytic strategy to constrain parameters as much as possible until there is a reliable indication that model fit has worsened substantially. 3.3.2. Constraints across time point The next set of chi-square difference tests compared models in which all predictor paths were constrained across time point (in addition to the aforementioned constraints across diagnostic group) to models in which these paths were allowed to vary by time point. In other words, these analyses tested whether the predictive relationship between each of the emotion regulation and personality trait variables and positive affect differs in magnitude at a particular time point relative to the others. Chi-square difference tests comparing models in which all predictor paths were constrained across time point to models in which these paths were unconstrained were nonsignificant for all five imputed data sets (all ps > .087). This finding demonstrates that, against expectation, all of the emotion regulation and personality trait variables were not significantly stronger prospective predictors of positive affect at any one particular time point. As such, all constraints across group and time point were retained for the final PANAS model. 3.4. Tests constraining parameters-SAM 3.4.1. Constraints across diagnostic group For the SAM data, chi-square difference tests comparing models in which all predictor paths were constrained across diagnostic group to models in which these paths were unconstrained were not significant for all five imputed data sets (ps > .060). This pattern of findings can be taken to suggest that, consistent with expectation, all parameters are similar enough for the GSAD and NOSAD groups that constraining them does not result in significantly worse fit. Accordingly, we proceeded with all constraints across group in subsequent analyses. 3.4.2. Constraints across time point Constraining all parameters across time point in addition to the aforementioned constraints across group yielded nonsignificant chi-square difference tests for two of the imputed data sets (ps > .290), with significant results in the remaining three data sets (ps < .048). This set of findings can be interpreted to reflect an inability to state for certain that constraining these paths does not lead to substantial decrement in fit. Put differently, this pattern suggests that there may be differential relationships between the predictors and positive affect across the four time points, consistent with hypothesis. Please see below regarding interpretation of the final model, in which only the constraints across diagnostic group were retained in accordance with these results. 3.5. Final model interpretation 3.5.1. PANAS The resulting model in which all predictor paths were constrained both across group and across time point showed very good fit for the PANAS data (CFI = .99, TLI = .99, RMSEA = .05). Autoregressive parameters in which PANAS positive affect at each time point was regressed on positive affect at the previous time point significantly predicted positive affect at T2, T3, and T4 (ps < .001). Further, difficulties engaging in goal-directed behavior was a significant predictor of positive affect at all three time points (p = .032). This result indicates that higher scores on difficulties with goal-directed behavior prospectively predict lower positive affect as hypothesized. Subsequent to finding that difficulties with goal-directed behavior was the only significant predictor of positive affect, we examined the collective influence of the remaining predictors by testing a model in which these predictive paths were removed. That is, we were interested in determining if it is necessary for the nonsignificant predictive paths to be included in the model, with the alternative hypothesis that their inclusion provides a less parsimonious account of the data. Tests of this post hoc theory revealed that removing the nonsignificant paths did not result in significant decrement to fit for all five data sets (ps > .178). Please see Fig. 1 for a depiction of this model. It should be noted that because the size of the variance of positive affect was allowed to vary by group and time point (and the standard deviation of some predictors also varied), there are small differences in the numerical value of the standardized coefficient in each group at each time point. Because the unstandardized parameters could be constrained across both group and time point, these differences should not be meaningful. Multiple-group autoregressive model of PANAS positive affect as predicted by ... Fig. 1. Multiple-group autoregressive model of PANAS positive affect as predicted by difficulties engaging in goal-directed behavior. The remaining predictors were nonsignificant and thus these prediction paths are not modeled. GSAD = generalized social anxiety disorder group; NOSAD = no social anxiety disorder group. All covariances among potential predictor variables included. Prediction paths are fully standardized parameters. Because of different standard deviations across groups, there are small numerical differences between groups at each time point. Accordingly, the GSAD parameters are presented for all paths that are constrained across group. Error terms are not modeled in this figure for simplicity. *p < .05, **p < .01, ***p < .001. Figure options 3.5.2. SAM The final constrained model showed very good fit (CFI = .99, TLI = 1.00, RMSEA = .03). In line with the aforementioned models, autoregressive parameters were significant at all three time points (ps < .009). In contrast to the final PANAS model, none of the emotion regulation or personality trait variables were significant prospective predictors of positive affect as measured by the SAM (all ps > .127). Although it is partially consistent with hypothesis that the predictive paths differed significantly in magnitude across time point, the fact that none of the variables was a significant predictor of positive affect at any particular assessment is against expectation. 3.6. Tests of an interaction between relationship type and predictors Finally, we evaluated our hypothesis that there would be a differential impact of emotionally-salient interactions with a romantic partner on positive affect, which we theorized would be reflected in an interaction between relationship type and each of the predictors. To this end, we tested models similar to the ones above, except that group was used as a predictor and relationship type served as the grouping variable. Further, all constraints across time point were retained. For both the PANAS and the SAM data, all five chi-square difference tests comparing models in which all predictive paths were unconstrained across relationship type to models in which these paths were constrained yielded nonsignificant results (all ps > .106). These findings indicate that, against expectation, the relationships between the predictors and positive affect did not differ between romantic partner relationships and friendships.