Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2024 Feb 6.
Published in final edited form as: Behav Ther. 2023 Jan 2;54(3):572–583. doi: 10.1016/j.beth.2022.12.009

Examining safety behavior subtypes across distinct social contexts in social anxiety disorder and major depression

Madeleine Rassaby a, Taylor Smith b, Charles T Taylor a,b
PMCID: PMC10846675  NIHMSID: NIHMS1951421  PMID: 37088511

Abstract

People with social anxiety disorder (SAD) use different types of safety behaviors that have been classified as avoidance vs. impression management. The current study investigated differences in safety behavior subtype use in 132 individuals with principal diagnoses of social anxiety disorder (SAD, n=69), major depressive disorder (MDD, n=30), and non-patient controls (n=33) across two social contexts: an interpersonal relationship-building task (social affiliation) and a speech task (social performance). We examined whether diagnostic groups differed in safety behavior subtype use and whether group differences varied by social context. We also explored relationships between avoidance and impression management safety behaviors, respectively, and positive and negative valence affective and behavioral outcomes within the social affiliation and social performance contexts. Safety behavior use varied by diagnosis (SAD > MDD > non-patient controls). The effect of diagnosis on impression management safety behavior use depended on social context: use was comparable for the principal SAD and MDD groups in the social performance context, whereas the SAD group used more impression management safety behaviors than the MDD group in the social affiliation context. Greater use of avoidance safety behaviors related to higher negative affect and anxious behaviors, and lower positive affect and approach behaviors across contexts. Impression management safety behaviors were most strongly associated with higher positive affect and approach behaviors within the social performance context. These findings underscore the potential value of assessing safety behavior subtypes across different contexts and within major depression, in addition to SAD.

Keywords: Safety Behaviors, Social Anxiety Disorder, Depression, Interpersonal, Social Performance

Introduction

Social anxiety disorder (SAD) is characterized by persistent fear of embarrassment or negative evaluation by other people, resulting in functional impairment and distress (American Psychiatric Association, 2013). Safety behaviors, defined as actions intended to prevent, minimize, or escape a feared outcome, are recognized as an important maintaining factor in SAD (Kirk et al., 2019; Piccirillo et al., 2016). Research suggests safety behaviors can be grouped into two main subtypes: avoidance and impression management (Clark & Wells, 1995; Evans et al., 2021; Gray et al., 2019; Plasencia et al., 2011). Avoidance safety behaviors serve to limit an individual’s involvement in a social situation or hide oneself (e.g., avoiding eye contact, minimizing talking, low self-disclosure), whereas impression management safety behaviors are intended to control the impression one makes on others in an effort to present a positive image (e.g., excessive rehearsal of conversation, self-monitoring). No research to date has examined the differential use of avoidance and impression management safety behaviors across more than one context (e.g., social affiliation versus social performance) in the same sample. Further, research has yet to explore safety behavior subtype use in disorders other than SAD (e.g., major depression). Finally, initial evidence suggests avoidance and impression management safety behaviors may be linked to different affective, behavioral, and social outcomes (Evans et al., 2021; Gray et al., 2019; Hirsch et al., 2004; Plasencia et al., 2011). Research is needed to replicate these findings and determine whether they extend across different social contexts. Identifying the presence and correlates of safety behavior subtypes across different contexts and disorders may inform a more precise understanding of which safety behaviors, in what contexts, and for whom, should be targeted in treatment.

Safety Behavior Subtypes and Social Context

Safety behaviors are strategic in that they are used in response to the demands and fears associated with a situation (Moscovitch et al., 2013). Notably, the demands and concerns evoked by one situation (e.g., a social interaction) may differ from those of another situation (e.g., a speech or other performance task). Different skills may be required to successfully engage in each context, and fears, distress levels, and coping strategies may vary. For example, social interactions require dynamic flow and reciprocal exchanges between two partners (e.g., attending and being responsive to one’s partner while thinking of things to say), while social performances increase focal attention on the performing individual, which can magnify the salience of being observed and potential conspicuousness of anxiety. Examining variability in safety behavior subtypes across different social contexts within the same sample has, to our knowledge, not been done. The present study sought to address this issue.

Safety Behaviors and Depression

Although research consistently implicates safety behaviors in the maintenance of SAD (Blakey & Abramowitz, 2016), less is known about their relevance to other conditions, such as major depressive disorder (MDD). This may be important given the high comorbidity rates between MDD and SAD (Kessler et al., 1999) and the fact that social impairment is common across both conditions (Robyn et al., 2020). Consistent with work in SAD (Plasencia et al., 2016; Taylor & Alden, 2011), if safety behaviors are used by those diagnosed with MDD, they may limit prosocial behaviors that facilitate social connection or diminish rewarding experiences from social engagements – factors related to the development and maintenance of depression (Kupferberg et al., 2016). In clinical practice, providers typically treat the principal, or most interfering diagnosis (Barlow et al., 2017). First-line treatments for MDD do not routinely assess for or target safety behaviors. If individuals with principal MDD engage in safety behaviors in social contexts, this information may be relevant to treatment planning. Initial evidence within SAD samples suggests higher depressive symptoms were related to increased safety behavior use (Plasencia et al., 2011; Rowa et al., 2015). To our knowledge, research has yet to examine whether individuals with principal major depression engage in safety behaviors (and which types) in the context of social situations. We aimed to address this gap by including individuals with principal MDD in this study.

Affective and Behavioral Correlates of Safety Behavior Subtypes

Insights into the unique correlates and consequences of avoidance and impression management safety behaviors have been demonstrated using a social interaction task in which an individual becomes acquainted with a stranger. In one such study, Plasencia and colleagues (2011) found avoidance strategies were positively related to state anxiety and to negative reactions from participants’ interaction partners. Impression management safety behaviors were not associated with state anxiety or partner liking, but were shown to impede corrections in negative predictions about future interactions. Gray et al. (2019) found healthy individuals who engaged in either avoidance or impression management safety behaviors experienced heightened anxiety in a social interaction task. Notably, avoidance safety behaviors had broader negative effects on the other individual in the conversation, such as liking their partner less and enjoying the conversation less, which were absent for the impression management safety behavior subtype. Taken together, these studies suggest avoidance and impression management safety behavior subtypes may have differential consequences in social interactions.

We are not aware of any studies that investigated safety behavior subtypes in social performance contexts, such as speech tasks. Studies exploring safety behaviors more broadly in speech tasks found safety behavior use is related to poorer performance and greater post-event processing (i.e., detailed review of a prior social situation; Mitchell & Schmidt, 2014; Rowa et al., 2015). Research is needed to explore whether avoidance and impression management safety behaviors have different correlates in performance-based contexts, like that observed in interpersonal contexts. Moreover, it is unknown whether avoidance and impression management safety behaviors elicit different emotional, behavioral, or social outcomes based on social context. For example, a behavior that may be helpful or benign in a performance situation may be more costly in a social interaction (e.g., brief mental rehearsal of what an individual is going to say during a speech may minimally interrupt performance whereas it may prevent the person from responding appropriately to an interaction partner because they missed part of what the other person said). Some research indicates judicious use of safety behaviors during the early stages of exposure therapy may not be detrimental to social performance, disconfirmatory learning, or affective state (Tutino et al., 2020; but see Rowa et al., 2015). Determining whether the correlates or outcomes of different safety behaviors vary across social contexts could inform where and when they should be targeted in treatment. The second aim of this study was to explore the relationships between avoidance and impression management safety behaviors, respectively, and positive and negative valence affective and behavioral outcomes across each social context.

Current Study

The goal of the present study was to investigate differences in safety behavior subtype use in individuals with principal diagnoses of SAD, MDD, and non-patient controls across two social contexts: an interpersonal relationship-building task (social affiliation) and a speech task (social performance). We aimed to examine whether diagnostic groups differed in safety behavior subtype use (avoidance and impression management; Aim 1) and whether group differences varied by social affiliation versus social performance contexts (moderator; Aim 2). We also sought to explore the affective and behavioral correlates of avoidance and impression management safety behaviors within each context (Aim 3). Data were obtained from baseline assessments conducted within clinical trials for SAD (NCT02136212) and MDD (NCT02330744). The research questions examined herein were not part of the parent trial aims.

We hypothesized that individuals with SAD would engage in the most safety behaviors (both avoidance and impression management) irrespective of social context (Aim 1), followed by individuals with MDD and then non-patient controls (i.e., main effect of diagnostic group). In the absence of past research, we did not make specific predictions about whether social context would moderate the hypothesized diagnostic group differences in use of safety behavior subtypes. For the exploratory correlation analyses, we predicted that avoidance safety behaviors would be associated with greater negative valence outcomes (state anxiety, negative affect, and anxious behaviors) and lower positive valence outcomes (positive affect, approach behaviors). We did not make predictions about the correlates of impression management safety behaviors due to mixed findings in previous studies. Finally, we did not make predictions about whether or how these relationships varied across contexts due to lack of prior research.

Method

Participants

The sample included 132 participants ages 18 to 39 years (M =24.05, SD = 4.67): 69 individuals (42 women, 26 men, one who identified as other) with a principal DSM-5 diagnosis of social anxiety disorder as determined by the Mini International Neuropsychiatric Interview (MINI) 7.0.01 or the Structured Clinical Interview for DSM-5 (SCID-5), 30 individuals (18 women, 12 men) with a principal DSM-5 diagnosis of major depressive disorder (MDD) according to the MINI, and 33 non-patient controls (23 women, 10 men) without a history of psychiatric diagnosis as determined by the MINI. Demographic and clinical characteristics are displayed in Table 1. Participants were recruited from the community through IRB-approved advertisements posted throughout community settings, online media, social media, and primary care clinics. This study represents a secondary analysis of baseline data obtained within the context of overarching treatment studies for SAD and MDD. ClinicalTrials.gov Identifiers: NCT02136212; NCT02330744.

Table 1.

Sociodemographic Characteristics of Participants

SAD MDD HC
Measure n % n % n %
Gender
 Female 42 60.9% 18 60% 23 69.7%
 Male 26 37.7% 12 40% 10 30.3%
 Other 1 1.4% 0 0% 0 0%
Ethnicity
 Hispanic 17 24.6% 9 31.0% 7 21.2%
 Not Hispanic 52 75.4% 20 69.0% 26 78.8%
Race
 Black 4 5.8% 3 10.3% 1 3.0%
 White 27 39.1% 10 34.5% 15 45.5%
 Asian-American 26 37.7% 4 13.8% 13 39.4%
 Native American 1 1.4% 0 0% 0 0%
 Native Hawaiian or Pacific Islander 2 2.9% 0 0% 0 0%
 Unknown or Declined to Respond 2 2.9% 2 6.9% 0 0%
 More than 1 race 5 7.2% 5 17.2% 4 12.1%
 Other 2 2.9% 5 17.2% 0 0%
Open in a new tab

Measures

Social anxiety symptoms

The Liebowitz Social Anxiety Scale (LSAS; Liebowitz, 1987) was used to assess social anxiety symptoms. It assesses fear and avoidance of social interaction and performance situations. Respondents are asked to rate their fear and avoidance for each of 24 situations on a 4-point scale ranging from “none/never” to “severe/usually.” The total score is calculated by summing the scores on each item. The LSAS demonstrates strong psychometric properties (Heimberg et al., 1999). Current sample’s Cronbach’s α = .98.

Depressive symptoms

The Beck Depression Inventory-II (BDI-II; Beck, Steer, & Brown, 1996) was used to assess symptoms of depression during the past two weeks. The BDI-II consists of 21 self-report items that are multiple choice and scored on a scale from zero to three. Total scores are calculated by summing the items and range from 0 to 63. The BDI-II is considered a reliable, well-validated measure (Beck et al., 1996; Dozois, Dobson, & Ahnberg, 1998). Current sample’s Cronbach’s α = .98.

Safety behaviors

A modified version of the Safety Behaviour Questionnaire (SBQ; Clark et al., 1995) as implemented in prior research (Plasencia et al., 2011) was utilized to assess avoidance and impression management safety behaviors (items and scoring from Plasencia et al., 2011). The SBQ measures specific strategies used by individuals in an effort to prevent feared social outcomes. Participants rated how frequently they utilized each strategy in the social affiliation and performance tasks on a 9-point scale (0 = never, 8 = always). To clarify that items reflected safety behaviors rather than general behaviors, items were answered in reference to behaviors that were used to “make yourself feel safer or to try to prevent your feared outcome(s) from happening.” Examples for avoidance safety behavior items include “avoided talking about yourself” and “said little or nothing (talking as little as possible).” Examples of impression management safety behavior items include “tried to conceal your anxiety” and “acted very agreeable.” The impression management subscale consisted of 9 items, and the avoidance subscale included 6 items (see Supplemental Materials for all items on each scale). Mean total scores were calculated for each subscale to facilitate interpretation of findings (possible range 0 to 8). Current sample’s Cronbach’s α = .85 for avoidance (6 items) and .87 for impression management (9 items) safety behaviors.

State anxiety

The state anxiety subscale of the State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983) was employed to gauge participants’ present state of anxiety directly following the social affiliation and speech tasks. The STAI consists of 20 items describing transitory feelings (e.g., “I am tense”) that are rated on a four-point intensity scale from 1 (not at all) to 4 (very much so). Scores may range from 20 to 80, with higher scores indicating greater anxiety. This measure demonstrates high internal consistency (α = .92; Barnes et al., 2002). Current sample’s Cronbach’s α = .94.

Positive and Negative Affect

The Positive and Negative Affect Schedule (PANAS; Watson et al., 1988) was utilized to assess positive and negative affect following the social affiliation and speech tasks. The PANAS consists of 20 self-report items that are rated on a five-point scale indicating the extent to which an individual felt the specified emotions during the task (e.g., “excited,” “ashamed”). The positive and negative affect subscales demonstrate high internal consistency, respectively (α = .89, α = .85; Crawford & Henry, 2002). Current sample’s Cronbach’s α = .93 for the positive affect subscale and .89 for the negative affect subscale.

Observer-rated social behaviors

Anxious behaviors consisted of six items reflective of observable displays of anxiety (show signs of anxiety, tremble or shake, speak fluently/clearly, create uncomfortable pauses, appear tense or rigid, fidget; Taylor & Alden, 2011). Social approach behaviors included five items reflective of prosocial behaviors (talk openly about yourself, appear actively engaged, convey interest in your partner/audience, appear friendly, talkative; Taylor & Alden, 2011). The items were rated on a 7-point scale with the anchors of “not at all” and “very much.” Past research demonstrated adequate internal consistency (Cronbach’s α range from .75-.90 and .90-.91, respectively; Taylor & Alden, 2011). The intraclass correlation coefficient (ICC; two-way mixed model) was used to assess interrater reliability. On the social affiliation task, the average measure ICC was .704 for approach behaviors (95% CI [.558, .807]) and .728 for anxious behaviors (95% CI [.595, .823]). For the speech task, the average measure ICC was .816 for approach behaviors (95% CI [.692, .885]) and .822 for anxious behaviors (95% CI [.737, .880]), indicating acceptable inter-rater agreement.

Design and Procedure

Study procedures were approved by the institutional review board and all participants provided informed written consent prior to engaging in study procedures. The current study involved completing questionnaires assessing social anxiety and depression levels and the completion of two behavioral approach tests, each of which was followed by a questionnaire on the utilization of safety behaviors during the test. The social affiliation task was completed first, followed by a filler task (to prevent carryover effects), then the social performance task. Participants were informed that the behavioral approach tasks would be video recorded. Upon completion, participants were debriefed.

Social affiliation task

Participants completed a social affiliation task with a trained confederate (for full task description and psychometric properties, see Hoffman et al., 2021). Participants and confederates alternated responding to a series of six questions with each question gradually increasing in intimacy level and self-disclosure elicited (see Supplemental Materials for the list of questions). Confederates were trained to engage in friendly, warm behavior.2 With the confederate present, the experimenter explained to the participant that the purpose of the task was to get to know each other by answering questions about themselves. Before leaving the room, the experimenter started a video recording of the dyad. The interaction lasted approximately 25 minutes. Hoffman and colleagues (2021) provided psychometric support for this paradigm as a reliable method of inducing and measuring social affiliation.

Filler task

Participants watched a five-minute neutral video of fish swimming to the sound of calming music.

Social performance task

Participants were asked to complete a five-minute impromptu speech (performance-based task), which is a common and well-established behavioral approach test used to assess anxiety and avoidance responses (Hofmann et al., 1995). The experimenter informed participants they would be video recorded for later quality assurance. Participants chose a controversial speech topic from a list of 5 options: abortion, nuclear power, corporal punishment, seatbelt laws, and the American health system. They were given two-minutes to prepare their speech, then given instructions to stand in a designated area in front of a video camera to deliver the speech. Participants were encouraged to speak for the full five minutes, but were told they could end the speech at any time by informing the experimenter.

Data Analytic Plan

Statistical analyses were conducted in SPSS version 25.0. Group differences on demographic and clinical characteristics were examined using chi-square tests for categorical variables and one-way analyses of variance (ANOVAs) for continuous variables. A 3 (Group: SAD, MDD, non-patient controls) × 2 (Context: social affiliation vs. social performance) repeated measures multivariate analysis of variance (MANOVA) was performed to examine whether the amount of impression management and avoidance safety behaviors varied based on diagnosis (between-subjects; Aim 1) and whether social context (within-subjects) moderated diagnostic group differences (Aim 2). Dependent variables were scores on the avoidance and impression management subscales of the SBQ – entered together within a MANOVA to protect against type I error inflation. We confirmed whether the data violated assumptions of MANOVA. Pillai’s trace was used when Box’s Test for Equivalence of Covariance Matrices was violated (Nimon, 2012). Univariate outcomes for each safety behavior subtype were examined following a significant (p < .05) multivariate effect. Pending significant main and/or interaction effects, Tukey’s HSD post-hoc analyses were conducted to compare groups on the relevant outcomes, and paired-samples t-tests were run to examine the differential use of safety behaviors across social contexts within each diagnosis separately. For the exploratory correlation analysis (Aim 3), bivariate correlations were computed in the combined SAD and MDD groups to investigate associations between safety behavior subtypes and outcome measures, including observer-rated anxiety and approach behaviors, positive and negative affect, and state anxiety following the social affiliation and social performance tasks, respectively. The clinical samples were combined for the correlational analyses in order to increase power.

Results

Preliminary Analyses

Table 2 summarizes group differences on social anxiety and depression symptom measures. One-way ANOVAs revealed a significant effect of diagnostic group on LSAS scores (SAD > MDD > non-patient controls; F(2, 129) = 192.68, p < .001, η2 = .53) and BDI-II scores (MDD > SAD > non-patient controls; F(2, 127) = 63.61, p < .001, η2 = .86) for the three conditions. There were no significant diagnostic group differences in gender, race, or ethnicity (all p > .05). A one-way ANOVA indicated a significant diagnostic group difference in age (F(2, 129) = 3.72, p = .027, η2 = .05). Specifically, a post hoc Tukey test showed that the principal MDD group (M = 25.57, SD = 5.24) was significantly older on average than the principal SAD group (M = 23.04, SD = 4.42; d = .52); there were no differences between the non-patient control group and the other groups in age.3

Table 2.

Means and Standard Deviations of LSAS and BDI-II Measures

SAD MDD HC F(2, 129) η2
Measure M SD M SD M SD
LSAS 82.48a 16.39 55.77b 26.95 8.70c 6.66 192.68*** .53
BDI-II 19.36a 10.59 25.68b 8.56 2.03c 2.52 63.61*** .86
Open in a new tab

Note.

***

p < .001.

Means with different subscripts (a, b, and c) indicate group differences at p < .05.

Primary Analyses

Does Safety Behavior Subtype Use Differ Across SAD, MDD, and Control Groups?

A 3 × 2 repeated measures MANOVA revealed safety behavior use differed significantly across the three diagnostic groups (Pillai’s Trace = 0.53, F(4,258) = 23.48, p < .001, η2 = 0.27), but did not differ across contexts (Pillai’s Trace = 0.01, F(2,128) = .61, p = .543, η2 = 0.01). Univariate tests revealed group differences for both impression management (F(2, 129) = 35.05, p < .001, η2 = .35) and avoidance safety behaviors (F(2,129) = 52.02, p < .001, η2 = .45). Tukey’s post-hoc comparisons indicated that the SAD group (M = 4.49, SD = 1.27) and MDD group (M = 4.09, SD = 1.49) did not differ on mean impression management safety behavior use (p = .333; d = .29), but they both engaged in more impression management safety behaviors than the non-patient controls group (M = 2.26, SD = 1.48) at p < .001 (SAD versus HC: d = 1.62; MDD versus HC: d = 1.24). Tukey’s post-hoc comparisons showed that the SAD group (M = 3.27, SD = 1.58) used significantly more avoidance safety behaviors on average than the MDD group (M = 2.17, SD = 1.48; d = .72), and both groups used significantly more avoidance safety behaviors than the non-patient controls group (M = 0.57, SD = 0.77) at p < .001 (SAD versus HC: d = 2.18; MDD versus HC: d = 1.36).

Does Social Context Influence Safety Behavior Subtype Use Across Diagnoses?

The main multivariate effect of diagnostic group was qualified by a significant group by social context interaction (Pillai’s Trace = .17, F(4,258) = 5.92, p < .001, η2 = 0.08). This multivariate effect was driven by impression management safety behaviors (F(2,129) = 12.06, p < .001, η2 = 0.16), but not avoidance safety behaviors (F(2,129) = 1.65, p = .196, η2 = .03; see Table 3 and previously reported post hoc tests for the main univariate effect of diagnosis on avoidance safety behaviors).

Table 3.

Means and Standard Deviations of Impression Management and Avoidance Safety Behaviors on Social Affiliation and Social Performance Tasks

SAD MDD HC
SB Type Context M SD M SD M SD
IM SBs Soc Affil 4.71a 1.19 4.03b 1.55 1.95c 1.30
Speech 4.27a 1.34 4.16a 1.42 2.57b 1.66
Av SBs Soc Affil 3.32a 1.51 2.19b 1.50 0.39c 0.65
Speech 3.22a 1.65 2.14b 1.46 0.74c 0.88
Open in a new tab

Note. Means with different subscripts (a, b, and c) indicate group differences at p < .05 on each respective task. IM SBs = impression management safety behaviors; Av SBs = avoidance safety behaviors; Soc Affil = social affiliation task; Speech = social performance task.

Impression Management Safety Behaviors by Social Context

Following the significant group by context interaction, results of a paired sample t-test indicated that the SAD group engaged in significantly more impression management safety behaviors on the social affiliation task (M = 4.71, SD = 1.19) compared to the social performance task (M = 4.27, SD = 1.34; p < .001; t(68) = 3.24, p = .002, d = .39). The MDD group, however, did not differ in impression management safety behavior use on the social affiliation task (M = 4.03, SD = 1.55) compared to the social performance task (M = 4.16, SD = 1.42; p = .426; t(29) = −0.81, p = .426, d = −.15). The non-patient controls group engaged in significantly more impression management safety behaviors on the social performance task (M = 2.57, SD = 1.66) compared to the social affiliation task (M = 1.95, SD =1.30; p < .001; t(32) = −3.65, p < .001, d = 0.64).

A one-way ANOVA revealed a significant effect of diagnostic group on impression management safety behavior use in the social affiliation context (SAD > MDD > non-patient controls; F(2, 129) = 50.18, p < .001, η2 = .44) and the social performance context (SAD = MDD > non-patient controls; F(2, 129) = 16.60, p < .001, η2 = .21). For the social affiliation task, the SAD group (M = 4.71, SD = 1.19) used significantly more impression management safety behaviors than the MDD group (M = 4.03, SD = 1.55; d = .49) and non-patient controls group (M = 1.95, SD =1.30; d = 2.21); the MDD group engaged in significantly more impression management safety behaviors than the non-patient controls group (d = 1.45). For the social performance task, the SAD group (M = 4.27, SD = 1.34) did not differ significantly in impression management safety behavior use than the MDD group (M = 4.16, SD = 1.42; d = .08); but both groups used more than the non-patient controls (M = 2.57, SD = 1.66; SAD versus HC: d = 1.13; MDD versus HC: d = 1.03). Table 3 and Figure 1 summarize group differences in impression management safety behavior use in each context.

Figure 1.

Figure 1

Open in a new tab

Average Impression Management Safety Behaviors Across Social Contexts by Diagnosis.

Note. Error bars indicate +/− one standard deviation.

Secondary Analyses

To assess whether findings based on principal diagnoses were influenced by comorbid diagnosis, we created a comorbid SAD and MDD group (n = 29), SAD only group (n = 50), MDD only group (n = 19), and non-patient controls group (n = 33), and repeated the analyses (see Supplemental Results for details). The pattern of findings for the SAD only group was consistent with those of the primary analyses for the SAD principal diagnosis group. Findings for non-patient controls also remained consistent. In contrast to the initial findings for the MDD principal diagnosis group, the MDD only group engaged in more impression management safety behaviors on the social performance task (M = 4.14, SD = 1.57) compared to the social affiliation task (M = 3.76, SD = 1.64; t(19) = −2.23, p = .038, d = −.50). The comorbid SAD and MDD group did not differ in impression management safety behavior use across contexts (social affiliation: M = 4.97, SD = 1.05; social performance: M = 4.69, SD = 1.14; t(28) = 1.46, p = .156, d = .27); this finding converges with that observed for the principal diagnosis MDD group in the main analyses.

A one-way ANOVA revealed a significant effect of diagnostic group on the number of impression management safety behaviors used in the social affiliation context (SAD+MDD = SAD; SAD+MDD > MDD; SAD = MDD; all patient groups > non-patient controls; F(3, 129) = 35.86, p < .001, η2 = .46) and the social performance context (SAD+MDD = SAD = MDD > non-patient controls; F(2, 129) = 12.73, p < .001, η2 = .23).

Do Safety Behavior Subtypes Relate to Affective and Behavioral Processes within Different Social Contexts?

Correlational Analyses – Social Affiliation Task

See Table 4 for intercorrelations among measures in the clinical sample for the social affiliation task (N = 99).4 Avoidance safety behaviors were positively associated with observer-rated anxious behaviors, and negatively associated with observer-rated approach behaviors. Avoidance safety behaviors were also positively correlated with self-reported negative affect and state anxiety, and negatively correlated with self-reported positive affect.

Table 4.

Intercorrelations among measures in clinical sample – Social Affiliation Task

Measure n M SD 1. 2. 3. 4. 5. 6 7.
1. Imp_SBs 99 4.51 1.34 --
2. Avoid_SBs 99 2.98 1.59 .26* --
3. Anx_Behav 99 19.99 4.74 .11 .31** --
4. App_Behav 99 23.07 4.36 −.03 −.39*** −.64*** --
5. STAI-S 99 46.28 9.46 .19 .45*** .36*** −.36*** --
6. PANAS_N 99 15.15 4.76 .25* .46*** .27** −.23* .68*** --
7. PANAS_P 99 20.72 6.83 .13 −.28** −.24* .33** −.52*** −.10 --
Open in a new tab

Note.

*

p < .05,

**

p < .01,

***

p < .001.

Imp_SBs = impression management safety behaviors (SBQ); Avoid_SBs = avoidance safety behaviors (SBQ); Anx_Behav = observer-reported anxious behaviors (SJQ); App_Behav = observer-reported approach behaviors (SJQ); STAI = state anxiety (STAI); PANAS_N = negative affect (PANAS); PANAS_P = positive affect (PANAS)

Impression management safety behaviors were positively correlated with self-reported negative affect; however, there were no significant correlations between impression management safety behaviors and observer-rated anxious or approach behaviors, or self-reported state anxiety or positive affect.

Correlational Analyses – Speech Task

Table 5 displays the intercorrelations among measures in the clinical sample for the speech task. Greater use of avoidance safety behaviors was associated with higher observer-rated anxious behaviors and lower observer-rated approach behaviors. Avoidance safety behaviors were also correlated with higher negative affect and state anxiety, and lower positive affect. Greater use of impression management safety behaviors was associated with lower observer-rated anxious behaviors, and higher observer-rated approach behaviors and self-reported positive affect.

Table 5.

Intercorrelations among measures in clinical sample – Social Performance Task

Measure n M SD 1. 2. 3. 4. 5. 6 7.
1. Imp_SBs 99 4.23 1.36 --
2. Avoid_SBs 99 2.89 1.66 .24* --
3. Anx_Behav 72 20.89 6.70 −.29* .37** --
4. App_Behav 72 20.85 4.92 .37** −.31** −.73*** --
5. STAI-S 99 51.62 11.36 .01 .49*** .55*** −.42*** --
6. PANAS_N 99 18.82 7.32 .19 .44*** .43*** −.32*** .77*** --
7. PANAS_P 99 19.93 7.64 .34** −.29** −.45*** .52*** −.52*** −.15 --
Open in a new tab

Note.

*

p < .05,

**

p < .01,

***

p < .001. Imp_SBs = impression management safety behaviors (SBQ); Avoid_SBs = avoidance safety behaviors (SBQ); Anx_Behav = observer-reported anxious behaviors (SJQ); App_Behav = observer-reported approach behaviors (SJQ); STAI = state anxiety (STAI); PANAS_N = negative affect (PANAS); PANAS_P = positive affect (PANAS)

Discussion

The current study explored avoidance and impression management safety behavior subtypes in individuals with SAD, MDD, and non-patient controls across social affiliation and social performance contexts. Safety behavior use varied by diagnosis, an effect that depended on social context for impression management but not avoidance safety behaviors. Further, different patterns of correlates emerged across safety behavior subtypes depending on context. These findings underscore the potential value of assessing safety behaviors across different contexts and within major depression, in addition to SAD, and may have implications for developing more targeted treatment approaches.

Social context moderated the relationship between diagnosis and impression management safety behavior use with a large effect size. Individuals with principal SAD engaged in more impression management safety behaviors on the social affiliation task compared to the social performance task with a small to medium effect size. In contrast, individuals with principal MDD did not differ in impression management safety behavior use across social contexts, and non-patient controls demonstrated greater use on the social performance task (medium effect size). We observed that in the social affiliation context, the principal SAD group engaged in more impression management safety behaviors than the principal MDD group, who engaged in more than the non-patient controls. However, in the social performance context, the SAD and MDD groups did not differ in impression management safety behavior use, but both engaged in more than the non-patient controls. These findings suggest individuals with principal depression may experience similar concerns to those with SAD about managing the impression they make on others in social situations, particularly in performance contexts.

When the primary analyses were repeated with comorbid versus non-comorbid diagnostic groups, results were mostly consistent, with the exception that the MDD group engaged in more impression management safety behaviors on the social performance task compared to the social affiliation task. The comorbid SAD and MDD group did not differ in impression management safety behavior use across contexts, suggesting the presence of both high social anxiety and depression symptoms influences impression management safety behavior use by reducing the differentiation between social contexts. Taken together, these results underscore the heightened use of impression management safety behaviors in both SAD and MDD, and also suggest social interaction contexts may induce particularly heightened impression management safety behavior use in individuals with SAD.

The SAD group used more avoidance safety behaviors than the MDD and control groups, irrespective of social context. The MDD group also engaged in more avoidance safety behaviors than the non-patient controls. These findings may indicate individuals with principal MDD experience heightened avoidance in social situations, though less so than those with principal SAD. This aligns with the findings by Plasencia and colleagues (2011) demonstrating that greater use of avoidance safety behaviors was associated with increased depressive symptoms in individuals with SAD.

Within the combined clinical samples (i.e., SAD and MDD), greater use of avoidance safety behaviors was related to negative valence outcomes, including higher negative affect (including anxiety), and greater observer-rated anxious behaviors across social contexts. Greater use of avoidance safety behaviors was also associated with lower positive affect and observer-rated approach behaviors across both social contexts. All effect sizes were medium-to-large in magnitude. These results replicate what was previously observed in interpersonal contexts (Evans et al., 2021; Gray et al., 2019; Hirsch et al., 2004), and extend these patterns of association into a performance context, suggesting avoidance safety behaviors may have pervasive detrimental consequences.

On the social affiliation task, impression management safety behaviors were not related to observer-rated anxious or approach behaviors. This finding is consistent with the literature linking avoidance, but not impression management safety behaviors to negative perceptions by observers on social affiliation tasks and poorer quality of interactions (Evans et al., 2021; Gray et al., 2019; Hirsch et al., 2004). Greater use of impression management safety behaviors correlated with higher negative affect (small effect), but not state anxiety. These findings add to but do not further clarify the already mixed literature on the relationship between impression management safety behaviors and state anxiety (Gray et al., 2019; but see Plasencia et al., 2011). A surprising finding was that, on the speech task, impression management safety behaviors were negatively correlated with observer-rated anxious behaviors (small-to-medium effect), and positively associated with positive affect and observer-rated approach behaviors (moderate effect sizes). This suggests impression management safety behaviors may be beneficial for one’s emotional state and ability to actively engage in performance situations – at least when measured in a circumscribed and controlled experimental context. It is possible impression management safety behaviors fulfill their intended purpose in performance situations, but not in more reciprocal, dynamic social interaction contexts. Indeed, it has been suggested that strategies like rehearsing phrases or feigning friendliness (i.e., impression management behaviors) may increase likability (Piccirillo et al., 2016). Determining the boundary conditions of such effects (e.g., in what contexts, at what frequency, and over how long are impression management displays helpful, benign, or costly) remains to be established.

Several limitations should be considered when interpreting results of this study. First, data on safety behavior utilization was collected via self-report. This may impact findings as participants could under- or overestimate their use of safety behaviors (e.g., due to social desirability, recall biases, or limited awareness). However, research indicates that self- and observer-report data of some safety behaviors correlate highly (Kocovsky et al., 2016). Second, the sample size for this study was relatively modest and was unbalanced across diagnostic groups, which may reduce power. Third, the MDD only group endorsed relatively elevated levels of social anxiety, which precludes forming conclusions about whether safety behavior usage in the MDD group reflects depression-specific processes or can be explained by the presence of subclinical levels of social anxiety. Future research may seek to disentangle the mechanisms underlying this relationship. Fourth, study tasks were not counterbalanced, which could lead to possible order effects (e.g., due to sensitizing or fatiguing effects from the first task). Fifth, exploratory analyses were correlational, preventing conclusions of causation. It is therefore possible that affective state may have impacted safety behavior use (rather than the reverse). Further, given that the clinical samples were combined to increase power for the correlational analyses, it was not determined whether findings differed by clinical diagnosis (SAD versus MDD). This study involved single social events, which may not be representative of longer-term outcomes of safety behavior use (e.g., threat disconfirmation). Finally, we did not assess other outcomes that may be impacted by impression management safety behaviors, such as maintaining threat predictions as was seen in Plasencia and colleagues (2011).

Despite limitations of this investigation, findings contribute to the body of literature on safety behaviors by providing insights into how social context and diagnosis may impact safety behavior use, and how safety behavior subtypes are differentially related to affective and behavioral outcomes across distinct contexts. Clinically, our results suggest it may be valuable to assess safety behavior use in people seeking treatment for principal MDD, in addition to SAD. Findings also indicate it may be valuable to consider safety behavior type and social context before deciding whether or when to address safety behaviors in exposure therapy. For example, clinicians may consider prioritizing addressing avoidance safety behaviors (before impression management safety behaviors) given that these are associated with more detrimental affective and behavioral outcomes. Considering the judicious use of some safety behaviors (e.g., impression management) within some contexts (e.g., social performance) could be considered early in treatment to facilitate engagement in especially challenging situations, after which safety behavior fading and elimination could proceed (Rachman et al., 2008; Goetz et al., 2016). It may also be useful to evaluate which contexts elicit which safety behaviors from a given individual in order to provide a more personalized or idiographic treatment approach. Further research on the longer-term impacts of safety behavior subtype use across different contexts could inform intervention targets and promote effective and efficient exposure-based therapy for individuals experiencing social anxiety and depression.

Supplementary Material

supplement

Highlights.

  • We examined safety behaviors (SBs) in social contexts (affiliation v performance).

  • SB subtype use varied by diagnosis (SAD > MDD > non-patient controls).

  • The effect of diagnosis on impression management SBs depended on social context.

  • SB subtype correlates depended on social context.

  • Avoidance SBs related to poorer emotional and behavioral outcomes.

Funding:

This research was supported by grants awarded to Charles T. Taylor from the National Institute of Mental Health (R00MH090243, R33MH113769), Brain and Behavior Foundation (21695), and the University of California, San Diego, National Institute of Health Clinical and Translational Science Awards Program Grant UL1TR001442.

Footnotes

Conflict of Interest: Charles T. Taylor declares that in the past 3 years he has been a paid consultant for Bionomics and receives payment for editorial work for UpToDate and the journal Depression and Anxiety. Madeleine Rassaby and Taylor Smith declare no conflicts of interest.

1

Because enrollment began prior to the release of MINI Version 7.0.0 for DSM-5, 10 participants were administered MINI Version 5.0.0 for DSM-IV.

2

Observer raters used a five-item scale assessing confederate warm / affiliative behavior displays (friendly, talkative, disinterested, distant, self-disclosive). Items were rated from 0 (not at all) to 7 (very much) and averaged to create a score. The mean warm / affiliative behavior display score was 5.87, with a standard deviation of .41.

3

No significant associations emerged between age and impression management or avoidance safety behaviors in either social context.

4

N = 72 for observer-reported anxious and approach behaviors on the social performance task. A sensitivity analysis was run including only subjects who had complete data for observer-reported behavior outcomes across both social contexts. The patterns of findings were similar to those reported in the main text. See Supplemental Materials for full analyses.

References

  1. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). [Google Scholar]
  2. Barnes LL, Harp D, & Jung WS (2002). Reliability generalization of scores on the Spielberger state-trait anxiety inventory. Educational and Psychological Measurement, 62(4), 603–618. 10.1177/0013164402062004005 [DOI] [Google Scholar]
  3. Barlow DH, Farchione TJ, Bullis JR, Gallagher MW, Murray-Latin H, Sauer-Zavala S, … & Cassiello-Robbins C (2017). The unified protocol for transdiagnostic treatment of emotional disorders compared with diagnosis-specific protocols for anxiety disorders: A randomized clinical trial. JAMA Psychiatry, 74(9), 875–884. doi: 10.1001/jamapsychiatry.2017.2164 [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Beck AT, Steer RA, & Brown GK (1996). Beck Depression Inventory (BDI-II) (Vol. 10, No. 3). London, UK: Pearson. [Google Scholar]
  5. Blakey SM, & Abramowitz JS (2016). The effects of safety behaviors during exposure therapy for anxiety: Critical analysis from an inhibitory learning perspective. Clinical Psychology Review, 49, 1–15. 10.1016/j.cpr.2016.07.002 [DOI] [PubMed] [Google Scholar]
  6. Clark DM, Butler G, Fennell M, Hackmann A, McManus F, & Wells A (1995). Social Behaviour Questionnaire. Unpublished manuscript.
  7. Clark DM, & Wells A (1995). A cognitive model of social phobia. In Liebowitz M, Hope DA, Schneier F, & Heimberg RG (Eds.), Social phobia: Diagnosis, assessment, and treatment. New York: Guildford Press. [Google Scholar]
  8. Crawford JR, & Henry JD (2004). The Positive and Negative Affect Schedule (PANAS): Construct validity, measurement properties and normative data in a large non‐clinical sample. British Journal of Clinical Psychology, 43(3), 245–265. 10.1348/0144665031752934 [DOI] [PubMed] [Google Scholar]
  9. Dozois DJA, Dobson KS, & Ahnberg JL (1998). A Psychometric Evaluation of the Beck Depression Inventory-II. Psychological Assessment, 10(2), 83–89. 10.1037/1040-3590.10.2.83 [DOI] [Google Scholar]
  10. Evans R, Chiu K, Clark DM, Waite P, & Leigh E (2021). Safety behaviours in social anxiety: An examination across adolescence. Behaviour Research and Therapy, 144, 103931. 10.1016/j.brat.2021.103931 [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Gray E, Beierl ET, & Clark DM (2019). Sub-types of safety behaviours and their effects on social anxiety disorder. Plos one, 14(10), e0223165. 10.1371/journal.pone.0223165 [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Goetz AR, Davine TP, Siwiec SG, & Lee HJ (2016). The functional value of preventive and restorative safety behaviors: A systematic review of the literature. Clinical psychology review, 44, 112–124. 10.1016/j.cpr.2015.12.005 [DOI] [PubMed] [Google Scholar]
  13. Heimberg RG, Horner KJ, Juster HR, Safren SA, Brown EJ, Schneier FR, & Liebowitz MR (1999). Psychometric properties of the Liebowitz Social Anxiety Scale. Psychological Medicine, 29(1), 199–212. 10.1017/S0033291798007879 [DOI] [PubMed] [Google Scholar]
  14. Hirsch C, Meynen T, & Clark D (2004). Negative self‐imagery in social anxiety contaminates social interactions. Memory, 12(4), 496–506. 10.1080/09658210444000106 [DOI] [PubMed] [Google Scholar]
  15. Hofmann SG (2007). Cognitive factors that maintain social anxiety disorder: A comprehensive model and its treatment implications. Cognitive Behaviour Therapy, 36(4), 193–209. 10.1080/16506070701421313 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hoffman SN, Thomas ML, Pearlstein SL, Kakaria S, Oveis C, Stein MB, & Taylor CT (2021). Psychometric evaluation of a controlled social affiliation paradigm: Findings from anxiety, depressive disorder, and healthy samples. Behavior Therapy, 52(6), 1464–1476. 10.1016/j.beth.2021.04.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kessler RC, Stang P, Wittchen HU, Stein M, & Walters EE (1999). Lifetime co-morbidities between social phobia and mood disorders in the US National Comorbidity Survey. Psychological Medicine, 29(3), 555–567. 10.1017/S0033291799008375 [DOI] [PubMed] [Google Scholar]
  18. Kirk A, Meyer JM, Whisman MA, Deacon BJ, & Arch JJ (2019). Safety behaviors, experiential avoidance, and anxiety: A path analysis approach. Journal of Anxiety Disorders, 64, 9–15. 10.1016/j.janxdis.2019.03.002 [DOI] [PubMed] [Google Scholar]
  19. Kocovski NL, MacKenzie MB, Albiani JJ et al. Safety Behaviors and Social Anxiety: An Examination of the Social Phobia Safety Behaviours Scale. J Psychopathol Behav Assess 38, 87–100 (2016). 10.1007/s10862-015-9498-6 [DOI] [Google Scholar]
  20. Kupferberg A, Bicks L, & Hasler G (2016). Social functioning in major depressive disorder. Neuroscience & Biobehavioral Reviews, 69, 313–332. 10.1016/j.neubiorev.2016.07.002 [DOI] [PubMed] [Google Scholar]
  21. Liebowitz MR, & Pharmacopsychiatry MP (1987). Social phobia. New York, NY: Guilford Publications. [Google Scholar]
  22. Mitchell MA, & Schmidt NB (2014). General in-situation safety behaviors are uniquely associated with post-event processing. Journal of Behavior Therapy and Experimental Psychiatry, 45(2), 229–233. 10.1016/j.jbtep.2013.11.001 [DOI] [PubMed] [Google Scholar]
  23. Moscovitch DA, Rowa K, Paulitzki JR, Ierullo MD, Chiang B, Antony MM, & McCabe RE (2013). Self-portrayal concerns and their relation to safety behaviors and negative affect in social anxiety disorder. Behaviour Research and Therapy, 51(8), 476–486. 10.1016/j.brat.2013.05.002 [DOI] [PubMed] [Google Scholar]
  24. Nimon KF (2012). Statistical assumptions of substantive analyses across the general linear model: a mini-review. Frontiers in psychology, 3, 322. 10.3389/fpsyg.2012.00322 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Piccirillo ML, Dryman MT, & Heimberg RG (2016). Safety behaviors in adults with social anxiety: Review and future directions. Behavior Therapy, 47(5), 675–687. 10.1016/j.beth.2015.11.005 [DOI] [PubMed] [Google Scholar]
  26. Plasencia ML, Alden LE, & Taylor CT (2011). Differential effects of safety behaviour subtypes in social anxiety disorder. Behaviour Research and Therapy, 49(10), 665–675. 10.1016/j.brat.2011.07.005 [DOI] [PubMed] [Google Scholar]
  27. Plasencia ML, Taylor CT, & Alden LE (2016). Unmasking one’s true self facilitates positive relational outcomes: Authenticity promotes social approach processes in social anxiety disorder. Clinical Psychological Science, 4(6), 1002–1014. 10.1177/216770261562220 [DOI] [Google Scholar]
  28. Rachman S, Radomsky AS, & Shafran R (2008). Safety behaviour: A reconsideration. Behaviour research and therapy, 46(2), 163–173. 10.1016/j.brat.2007.11.008 [DOI] [PubMed] [Google Scholar]
  29. Rowa K, Paulitzki JR, Ierullo MD, Chiang B, Antony MM, McCabe RE, & Moscovitch DA (2015). A false sense of security: Safety behaviors erode objective speech performance in individuals with social anxiety disorder. Behavior Therapy, 46(3), 304–314. 10.1016/j.beth.2014.11.004 [DOI] [PubMed] [Google Scholar]
  30. Spielberger CD (1983). State-trait anxiety inventory for adults. 10.1037/t06496-000 [DOI]
  31. Taylor CT, & Alden LE (2011). To see ourselves as others see us: an experimental integration of the intra and interpersonal consequences of self-protection in social anxiety disorder. Journal of Abnormal Psychology, 120(1), 129. 10.1037/a0022127 [DOI] [PubMed] [Google Scholar]
  32. Tutino JS, Ouimet AJ, & Ferguson RJ (2020). Exploring the impact of safety behaviour use on cognitive, psychophysiological, emotional and behavioural responses during a speech task. Behavioural and Cognitive Psychotherapy, 48(5), 557–571. 10.1017/S135246582000017X [DOI] [PubMed] [Google Scholar]
  33. Watson D, Clark LA, & Tellegen A (1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(6), 1063–1070. 10.1037/0022-3514.54.6.1063 [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS1951421-supplement-supplement.docx (40.6KB, docx)

RESOURCES