State and Trait Emotion Regulation Diversity in Social Anxiety

Abstract

Emotion regulation (ER) diversity, defined as the variety, frequency, and evenness of ER strategies used, may predict social anxiety severity. In a sample of individuals with high (n = 113) and low (n = 42) social anxiety severity, we tested whether four trait ER diversity metrics predicted group membership. We generalized existing trait ER diversity calculations to repeated measures data to test whether state-level metrics (using 2 weeks of ecological momentary assessment [EMA] data) predicted social anxiety severity within the higher severity group. As hypothesized, higher trait ER diversity within avoidance-oriented strategies predicted greater likelihood of belonging to the higher severity group. At the state level, higher diversity across all ER strategies, and within and between avoidance- and approach-oriented strategies, predicted higher social anxiety severity (but only after analyses controlled for number of submitted EMAs). Only diversity within avoidance-oriented strategies was significantly correlated across trait and state levels. Findings suggest that high avoidance-oriented ER diversity may co-occur with higher social anxiety severity.

Keywords

social anxiety emotion regulation emotion regulation diversity emotion regulation flexibility ecological momentary assessment

People with social anxiety disorder (SAD) have difficulty regulating their emotions (Jazaieri et al., 2015). Specifically, they tend to overly rely on strategies that distance themselves from their anxious thoughts and feelings (e.g., Kashdan et al., 2013) and conceal their negative emotions from others (e.g., Farmer & Kashdan, 2012). People with SAD also report a limited repertoire of strategies beyond those that are avoidant (Rusch et al., 2012). Taken together, people with SAD devote considerable effort to managing their emotions through primarily avoidance-oriented strategies (Kashdan et al., 2011), and persistent avoidance is core to the pathology (American Psychiatric Association, 2013).

Persistent avoidance in the relative absence of many other emotion regulation (ER) strategies has resulted in socially anxious people being described as “inflexible” and “rigid” regulators (Goodman & Kashdan, 2021). This rigidly avoidant regulation pattern fits with recent conceptualizations of emotion dysregulation that emphasize inflexible use of a few strategies regardless of changing situational demands and opportunities (Aldao et al., 2015; Bonanno & Burton, 2013). In the ER flexibility framework, no individual strategy is considered universally adaptive or maladaptive. Rather, this framework assumes that all strategies have the potential to be helpful under certain conditions, and the framework therefore predicts that healthy regulators show variety in the strategies they use across situations (Aldao et al., 2015). ER diversity measures the variety,¹ frequency,² and evenness³ of the ER strategies that a person uses, thereby capturing the variation in ER that is expected of a healthy regulator. Given SAD’s dysregulated and rigid clinical presentation, people with higher levels of social anxiety may typically show lower ER diversity. However, ER diversity does not consider how strategically matched the strategies used are to the situation at hand or how effectively those strategies are implemented. Therefore, people with higher levels of social anxiety may show higher ER diversity as a result of experiencing more persistent emotional distress (Kashdan et al., 2014), which may motivate more frequent, variable, and intense—albeit less effective—regulatory attempts. Thus, determining whether ER diversity is a reliable marker of social anxiety—and if so, in what direction—is a first step toward evaluating ER diversity as a future intervention target. Building from mathematical operationalizations of ER diversity offered by Wen and colleagues (Wen et al., 2021), we calculated four different measures of trait ER diversity and tested whether these trait diversity measures differentiate between people reporting high versus low levels of social anxiety.

Wen and colleagues’ trait ER diversity equations are an exciting contribution to the growing ER diversity and ER flexibility literatures; however, their equations were designed for ER data captured at a single time point. Given that sampling ER strategy use multiple times throughout a person’s daily life offers a clearer and more ecologically valid picture of individuals’ ER efforts over time, we extended Wen and colleagues’ (2021) trait equations to repeated measures data of the sort collected in ecological momentary assessment (EMA) studies. Specifically, participants who reported relatively higher and lower levels of social anxiety severity rated their typical (trait) use of ER strategies, and then those in the high-social-anxiety group also reported their in-the-moment ER strategy use in an EMA study for 2 weeks. We then tested (a) whether trait ER diversity predicts high- versus low-social-anxiety group membership, (b) whether state ER diversity predicts severity of self-reported social anxiety, and (c) the extent to which trait and state ER diversity scores are correlated. The second and third tests were conducted with the high-social-anxiety group, given that they were the only group to complete the EMA portion of the study and they provide a valuable opportunity to understand ER diversity among individuals who routinely experience emotion dysregulation.

ER Diversity in Social Anxiety

We identified competing hypotheses for the relationship between ER diversity and level of social anxiety symptoms (Hypothesis 1a). On the one hand, lower ER diversity might indicate a limited repertoire (e.g., not having access to a range of different strategies) or rigid use of just a few strategies (e.g., having access to many strategies but tending to use the same few over and over regardless of the situation at hand). Thus, lower ER diversity may be associated with higher (vs. lower) social anxiety severity, given that SAD is characterized by rigid overreliance on avoidance-oriented strategies (Kashdan et al., 2011).

On the other hand, the intolerance of negative emotions observed in SAD (Turk et al., 2005) may over time lead socially anxious people to use any possible strategy to try to make their anxiety and negative emotions go away as fast as possible. Indeed, despite social anxiety being associated with lower perceived availability of regulatory strategies (Rusch et al., 2012), people with SAD have been found to use more regulatory strategies per stressful event than do healthy control participants (Goodman et al., 2021). Further, although a different clinical population, currently depressed individuals reported higher ER diversity than did healthy control participants (Wen et al., 2021). Therefore, higher ER diversity might indicate more urgency, effort, and/or opportunity to reduce negative emotion. From this perspective, we might expect higher ER diversity to relate to higher (vs. lower) social anxiety severity.

Differentiating avoidance- from approach-oriented strategies in ER diversity

A person’s ER diversity score depends on how many strategies (out of all assessed strategies) they endorse using more frequently. ER diversity does not vary as a function of which specific strategies are endorsed or not endorsed (Wen et al., 2021). Recognizing that the association between ER diversity and social anxiety severity may depend not only on how many strategies a person endorses but also on which strategies they endorse using more frequently, we calculated two additional ER diversity metrics. These metrics explicitly account for whether each endorsed strategy orients a person toward their emotional experience (i.e., an approach-oriented strategy) or orients a person away from their emotional experience (i.e., an avoidance-oriented strategy). Notably, Wen and colleagues chose to categorize strategies into a group of putatively adaptive strategies and a group of putatively maladaptive strategies. Although the adaptive/maladaptive distinction they used was intuitively appealing and a logical starting point for their initial test of ER diversity, ER flexibility theory emphasizes the importance of choosing the strategy that is best able to serve the desired regulatory function a person holds in the moment. Therefore, rather than distinguish between strategies that are associated with better and worse outcomes, on average, we chose to conceptually organize strategies with reference to how they typically orient a person to the emotional experience that they are regulating (e.g., approaching the emotion via ruminating on it and its meaning vs. avoiding the emotion via distraction). We were interested in distinguishing avoidance- from approach-oriented strategies given the emphasis of approach- and avoidance-related processes in SAD (Rapee & Heimberg, 1997) and because previous factor analyses have found support for categorizing strategies as either engagement-oriented or avoidance-oriented (Daros et al., 2019; McMahon & Naragon-Gainey, 2019). We refer to diversity among approach-oriented strategies and diversity among avoidance-oriented strategies as within-factor diversity and diversity in using avoidance- versus approach-oriented strategies as between-factor diversity.

Within-factor diversity

Within-factor diversity measures the extent to which and evenness with which a person reports using different ER strategies within a particular grouping of strategies rather than out of all possible strategies. In this context, “extent” refers to how frequently each strategy is used, and “evenness” refers to how equally frequently all strategies are used relative to each other. Thus, within-factor diversity is calculated separately for each factor, or grouping, of strategies (i.e., avoidance- or approach-oriented strategies). So only strategies comprising a given factor are considered when calculating within-factor diversity.

We calculated within-avoidance diversity and within-approach diversity separately, yielding two within-factor diversity scores per person. We expected that higher within-avoidance diversity would predict increased likelihood of that person endorsing higher (vs. lower) social anxiety severity because socially anxious people tend to more frequently use avoidance-oriented strategies than less anxious people do (Hypothesis 1b; e.g., Farmer & Kashdan, 2012). Conversely, we expected that higher within-approach diversity would predict increased likelihood of that person endorsing lower (vs. higher) social anxiety severity because less anxious people typically endorse greater access to a range of ER strategies, especially approach-oriented strategies, than more anxious people do (Hypothesis 1c; Rusch et al., 2012). Although we focused our hypotheses on social anxiety, our predictions were consistent with Wen and colleagues’ (2021) finding that people with current and remitted depression showed greater diversity in maladaptive strategies (which are typically more avoidance-oriented) but less diversity in adaptive strategies (which are typically more approach-oriented) compared with healthy individuals.

Between-factor diversity

Rather than treat each strategy or factor separately, between-factor diversity calculates the extent to which and evenness with which a person reports using strategies across multiple factors of ER. Suppose two people reported using four out of eight assessed strategies extremely frequently and never used the remaining four strategies. Both people would receive the same overall ER diversity score. However, imagine that all four strategies the first person reported using were categorized within the avoidance-oriented factor, whereas the second person’s four endorsed strategies were equally divided between avoidance- and approach-oriented factors. In this case, despite both individuals receiving an identical overall ER diversity score, the first person would receive a lower between-factor diversity score than the person whose strategy endorsements spanned both factors. Because greater between-factor diversity reflects more even and frequent use of both avoidance- and approach-oriented strategies and using both types of strategies suggests greater ER flexibility, we expected that higher between-factor diversity would predict increased likelihood of that person endorsing lower (vs. higher) social anxiety severity (Hypothesis 1d). That said, we acknowledge that adaptive ER might not be characterized by equal use of avoidance- and approach-oriented strategies. Instead, relatively greater use of approach- relative to avoidance-oriented strategies might be more adaptive. However, in the absence of contextual information to speak to how much more often approach-oriented strategies should be used over avoidance-oriented strategies, we formed this hypothesis on the basis of the assumption that flexible regulation increases ER diversity.

State ER Diversity

As part of the interest around flexible ER, there has been a shift toward assessing ER in people’s day-to-day lives in addition to measuring self-reports of typical ER tendencies. Given that we collected state reports of ER strategy use within our high-social-anxiety participant group, we were in a unique position to investigate the relationship between severity of social anxiety and these ER diversity metrics when they were calculated on repeated in-the-moment (i.e., state) ER reports. All our hypotheses about the relation between state ER diversity metrics and continuously measured social anxiety severity are consistent with our hypotheses about the relation between trait ER diversity metrics and high- versus low-social-anxiety group membership. However, we recognize that by conducting these state ER diversity analyses in a group of participants experiencing a wide but elevated range of social anxiety, associations may be suppressed because of a restriction in range (Hypotheses 2a–2d).

Trait Versus State ER

Given that we had trait and state reports of ER strategy use within the same group of participants, we tested how strongly correlated aspects of ER diversity at a trait level were with those same aspects of ER diversity when derived from state reports within our participants scoring high on a measure of social anxiety. Because research suggests that trait and state ER reports do not necessarily correlate strongly (Blalock et al., 2016; Brockman et al., 2017), Analyses 3a to 3d were exploratory. We expected that these analyses would provide important insights into how self-concept about general ER strategy use matches with observed use in daily life.

Method

Procedure

As part of a larger parent data collection, we recruited participants who scored either relatively high or relatively low on a measure of social anxiety severity. Participants in the low-social-anxiety group were recruited from an undergraduate psychology participant pool, whereas participants in the high-social-anxiety group were recruited mainly from the same undergraduate psychology participant pool but also through university LISTSERV emails, community flyers, and online postings seeking “socially anxious individuals aged 18-45 to participate in a five-week smartphone monitoring study.” To determine eligibility, we screened prospective participants with the Social Interaction Anxiety Scale (SIAS), a 20-item self-report scale measuring social anxiety in one-on-one and group situations (Mattick & Clarke, 1998). Forty-two participants scoring relatively low on the measure (≤ 10 on the SIAS, which is three fourths of a standard deviation below the mean of a previous community sample; Mattick & Clarke, 1998) and 113 participants scoring relatively high on the same measure (29 or higher on the SIAS, which is approximately one fourth of a standard deviation below the mean in a sample diagnosed with SAD; Mattick & Clarke, 1998) provided written informed consent to participate in the study, which was approved by the University of Virginia Institutional Review Board (No. 2018-0018-00).⁴ Sample size was determined per the parent data collection. Specifically, we included all of the more highly socially anxious participants that we recruited for the intervention trial in the parent data collection (see Daniel, Daros, et al., 2020). We also recruited approximately as many individuals with low social anxiety as were in the high-social-anxiety control group to support a separate investigation (see Beltzer, 2022). SAD diagnosis status was not assessed, and participants in both the high- and low-social-anxiety groups were recruited simultaneously. Participants’ sex, race, ethnicity, and level of education did not significantly differ between the high- and low-social-anxiety groups (all ps > .196). However, participants in the high-social-anxiety group (mean age = 20.38) were significantly older than participants in the low-social-anxiety group (mean age = 19.38), t(115) = −2.50, p < .05. Participant demographics are presented in Table 1.

Table 1.

Participant Demographics

Variable	Trait analyses		State analyses
Variable	Low-social-anxiety group(n = 42)	High-social-anxiety group(n = 113)	High-social-anxiety group(n = 105)
Age (years)	M = 19.38, SD = 1.87	M = 20.38, SD = 2.95	M = 20.49, SD = 3.02
Gender (n)
Female	30 (71.43%)	85 (75.22%)	78 (74.29%)
Male	12 (28.57%)	28 (24.78%)	27 (25.71%)
Race (n)
White	28 (66.67%)	76 (67.26%)	72 (68.57%)
Asian	8 (19.05%)	18 (15.92%)	16 (15.24%)
Black	1 (2.38%)	9 (7.96%)	7 (6.67%)
Middle Eastern	1 (2.38%)	2 (1.77%)	2 (1.90%)
Multiracial	2 (4.76%)	8 (7.08%)	8 (7.62%)
Prefer not to answer	2 (4.76%)	0 (0.00%)	0 (0.00%)
Ethnic identity (n)
Latinx/Hispanic	2 (4.76%)	3 (2.65%)	3 (2.86%)
Not Latinx/Hispanic	40 (95.24%)	109 (96.46%)	101 (96.19%)
Preferred not to answer	0 (0%)	1 (0.88%)	1 (0.95%)
Education level (n)
Bachelors’ degree	41 (97.62%)	107 (94.69%)	99 (94.29%)
Master’s degree	1 (2.38%)	5 (4.42%)	5 (4.76%)
Doctoral degree	0 (0.00%)	1 (0.88%)	1 (0.95%)
Social Interaction Anxiety Scale
M (SD)	6.76 (3.01)	46.47 (10.31)	46.89 (10.08)
Range	0–10	29–73	29–73
Patient Health Questionnaire (eight items)
M (SD)	5.26 (4.48)	9.40 (5.62)	9.49 (5.61)
Range	0–16	0–24	0–24

Note: Trait-level analyses compared participants in the low-social-anxiety and high-social-anxiety groups. State-level analyses were conducted only on participants in the high-social-anxiety group. Demographic statistics differ between participants with high social anxiety in the trait versus state analyses because eight participants with high social anxiety did not submit any eligible ecological momentary assessment surveys and were subsequently excluded from state analyses.

All participants, regardless of social-anxiety-group membership, rated the extent to which they used 17 different ER strategies over the past 2 weeks. Participants belonging to the high-social-anxiety group were subsequently enrolled in a 5-week follow-up EMA study in which they reported whether they used 19 ER strategies (the 17 that were assessed at the trait level plus two additional strategies) in the 30 min leading up to randomly timed surveys deployed to their smartphone up to 6 times a day. In addition to reporting their state ER strategy use, participants also reported their state affect and anxiety, motivation to change their emotions, current location, current social context, and the extent to which they thought their regulation attempts were helpful. An intervention designed to shift negative interpretation bias was delivered randomly to half of the participants during Week 3 of this 5-week EMA study. Therefore, we used only the first 2 weeks of EMA self-reports because intervention-condition differences did not influence study procedures and participant reporting in the first 2 weeks. Participants completed 67.8% of all scheduled EMA surveys during these 2 weeks, with each participant completing an average of 56.9 surveys (SD = 22.04). Although this is a secondary data analysis, we have not yet analyzed these trait ER data, and we have not yet applied an ER diversity metric to any of these data (trait or state). The work described in the current article differs from all other work in that we used different methods of analysis and our hypotheses were tied to conceptually distinct questions.

Measures

Social anxiety

Social anxiety severity was measured dimensionally using the 20-item SIAS questionnaire (Mattick & Clarke, 1998). Participants rated their agreement with 20 statements on a Likert scale ranging from 0 (not at all characteristic of me) to 4 (extremely characteristic of me), with higher scores reflecting more severe social anxiety. Internal consistency was excellent in the current sample (α = .96).

Depression

Depression severity was measured dimensionally using the eight-item Patient Health Questionnaire (PHQ-8: Kroenke et al., 2009). Participants rated how frequently they had been bothered by each of eight different depression-related symptoms over the previous 2 weeks using a Likert scale ranging from 0 (Not at all) to 3 (Nearly every day). Higher scores reflect more severe depression. Internal consistency was good in the current sample (α = .87).

Trait ER frequency

Participants were presented with a list of 17 different ER strategies that described each strategy in layperson terms (see Table 2). Participants were asked to rate the “extent to which you’ve used each strategy to manage your emotions over the past two weeks” using a Likert scale ranging from 0 (not at all) to 6 (a lot). Strategies earning a higher score indicate more frequent use. This measure was adapted from the work of Aldao and Dixon-Gordon (2014) for the current study, and we selected strategies to broadly sample ER across cognitive, behavioral, and interpersonal domains. We described each strategy using language either from existing single-item strategy measures (i.e., Heiy & Cheavens, 2014) or that we developed in consultation with previous participants and undergraduate research assistants.

Table 2.

Conceptual Factor Structure of Emotion-Regulation Strategies Assessed in This Study

Factor and strategy
Approach-oriented
Tackling the issue head on
Seeking advice/comfort from others
Coming up with ideas/plan for action
Accepting thoughts/feelings
Thinking of the situation differently
Ruminating about something
Criticizing myself
Avoidance-oriented
Eating
Hiding my thoughts/feelings from others
TV/Internet/gaming
Sleeping
Ignoring/avoiding certain people/situations
Doing something fun with others
Pushing away bad thoughts
Thinking about things that went/are going well
Distracting myself^a
Exercising^a

Note: The statements given here are the layperson descriptions of the strategies that we presented to participants. ^aThis strategy was presented at the state level but not at the trait level.

State ER frequency

ER strategy use in daily life was measured using a check-all-that-apply list in response to the statement, “Over the 30 minutes before the survey prompt, I tried to change my thoughts and feelings through . . .” The prompt intentionally asks about responses to feelings broadly rather than limiting ER to only negative emotions, given that SAD is associated with difficulties managing both negative and positive emotions (Farmer et al., 2014). All strategies measured at the trait level were assessed in daily life using identical layperson language, with the addition of two other strategies (i.e., exercising and distraction). Each strategy was measured using a binary response scale (1 = strategy was endorsed, 0 = strategy was not endorsed). Participants could endorse multiple strategies within the same survey. Participants could also report that they were not currently regulating their emotions.

State affect

At each EMA survey, participants rated their momentary affect using the single item, “Right now, I am feeling . . .” Anchors ranged from 1 (very negative) to 10 (very positive).

Conceptually organizing ER strategies into factors

We organized each of the assessed ER strategies into an approach- versus avoidance-oriented factor structure based on how each strategy typically orients a person toward the emotional experience they are attempting to regulate. The factor groupings and layperson text shown to participants are shown in Table 2.

Analytic plans

All analytic plans—including our approach-versus-avoidance conceptual grouping of ER strategies—and hypotheses were preregistered on OSF and can be accessed at https://osf.io/vn8g4. Data and analysis scripts, including R functions built to calculate these trait and state ER diversity metrics, are also publicly available on our OSF project page (https://osf.io/xadyp/).

Data reduction

Although we measured alcohol use and drug use at both trait and state levels as part of the parent study, we decided a priori to remove these two strategies from these analyses because it is unlikely that using drugs or alcohol to regulate one’s emotions would be an adaptive option for regulating. Thus, this was a conceptual decision because ER diversity is theorized to promote enhanced well-being by allowing for more flexible use of many strategies to meet the specific demands of different situations more adaptively, reasoning which thereby assumes that all strategies included in the ER diversity calculation might routinely be helpful under certain circumstances. Although there are situations in which use of drugs and alcohol could be an adaptive regulatory strategy, we expected that those instances would be very rare in this sample and did not want to introduce the conceptual muddiness that would come from including them with the other assessed strategies. We were further confident in our choice to remove these strategies because they were endorsed extremely infrequently in the daily life data collected for the current study (< .5% of the time across all EMA survey responses). Thus, we investigated diversity across the 15 trait ER strategies and 17 state ER strategies described in Table 2.

The EMA data collection yielded 12,747 survey responses across 113 high socially anxious participants. Following our preregistered analytic approach, we removed all surveys that were not submitted within the first 2 weeks of the study, which left 6,432 surveys. To focus on the diversity of strategies used in daily life, assuming some form of ER was reported, we further removed observations in which participants reported not regulating their emotions by any means throughout the previous 30 min. This data-reduction choice was preregistered, and it reduced our final data set to 2,454 surveys across 105 high socially anxious participants.

Calculating ER diversity from continuous trait ER reports

Overall ER diversity

The formula we used to assess trait overall ER diversity was

- Σ_{i = 1}^{s} p_{i} \times 1 n (p_{i}),

where s is the total number of ER strategies and p_i is the proportion of the maximum possible score across all ER strategies made up of the ith ER strategy score. This equation is defined by Wen et al. (2021) and captures frequency as the proportion of each strategy’s use (with strategy-level proportions closer to 1 indicating more frequent use) and evenness as the extent to which each ER strategy’s proportion is similar across all strategies.

The overall ER diversity score reflects the extent to which and evenness with which a person reports using s number of ER strategies, regardless of how the strategies are conceptually related to each other. In other words, the diversity with which a person reports using all ER strategies without reference to the underlying approach- versus avoidance-oriented function of the strategies reported. In the current study, s = 15 and each strategy was measured on a Likert scale from 0 to 6, yielding a maximum⁵ p_i to be calculated out of 90 (15 × 6 = 90).

Within-factor diversity

An ER within-factor diversity score reflects the extent to which and evenness with which a person reports using s number of ER strategies within a given factor of strategies. It uses the same equation as is defined above. However, the values of s and p_i vary as a function of the specific factor under investigation. Specifically, for the approach-oriented factor, s = 7 and p_i is calculated out of a maximum 42 (7 × 6 = 42). For the avoidance-oriented factor, s = 8 and p_i is calculated out of a maximum 48 (8 × 6 = 48).

Between-factor diversity

The formula we used to assess trait between-factor diversity was

- \sum_{i = 1}^{k} p_{i} \times \ln (p_{i}),

where k is the total number of ER strategy factors and p_i is the proportion of the maximum possible score across all ER strategy factors made up of the ith ER strategy score. This equation is an extension of what was defined by Wen et al. (2021).

The ER between-factor diversity score reflects the extent to which and evenness with which a person reports using strategies across multiple factors of ER. In the current study, k = 2 (approach and avoidance) and p_i is still calculated out of a maximum 90, given that we organized each of the 15 strategies into one of the two factors.

Calculating ER diversity from repeated measures binary ER reports

Overall ER diversity

To create a vector of overall ER diversity scores of length i, where i is the total number of participants, we used the following formula:

- \sum_{j = 1}^{s} p_{j} \times \ln (p_{j}),

where j is the total number of repeated observations in which some form of ER was reported, s is the total number of ER strategies, and p_j is the proportion of the maximum possible score across all ER strategies across observations j. This equation is a repeated measures extension of what was defined by Wen et al. (2021).

The interpretation of this equation’s product is consistent with the trait overall ER diversity score described above. Here, the denominator for p_j is set to all eligible responses j for a given individual i multiplied by the maximum possible score across all s ER strategies for a given observation. In the current study, because strategy use was measured with an independent binary response scale per strategy, the maximum score for a given observation j is s. In state ER reports, overall s = 17.

Within-factor diversity: state level

As was the case at the trait level, the equation for deriving within-factor diversity at the state level is given by the overall ER state-level equation. However, the denominator for p_j was set to all eligible responses j for individual i multiplied by s within a given factor k (s_avoidance = 10, s_approach = 7). The interpretation of these products is consistent with the trait within-avoidance and trait within-approach diversity scores described above.

Between-factor diversity: state level

Again, the only difference between the overall ER and between-factor diversity equations is to replace s (17) with k (2). The product of this equation is interpreted consistently with the trait between-factor diversity score described above. Here, the denominator for p_j is set to all eligible responses j for individual i multiplied by s.

Note that if an individual never reported using a given strategy across all their survey responses, the p_ij term for that strategy was set to 0.

Models

We calculated overall ER diversity, between-factor diversity, and within-factor diversity according to the equations defined above using trait and state self-reported ER data. We used a base of 2 when taking the log to increase interpretability.

To test whether trait ER diversity metrics can differentiate between high- and low-social-anxiety groups (Hypotheses 1a–1d), we conducted four logistic regression analyses using the glm function from the stats package in R (R Core Team, 2017). In these analyses, each of the trait-level diversity scores (overall ER diversity, between-factor diversity, within-avoidance diversity, and within-approach diversity) were separately used to predict social anxiety group (1 = high SAD symptom group; 0 = low SAD symptom group).

To test the association between state ER diversity metrics and social anxiety severity (Hypotheses 2a–2d), we first conducted four regression-based models using the lm function in the stats package. Each state-level diversity metric was separately used to predict social anxiety severity. In these analyses, social anxiety severity was measured continuously with the SIAS, and scores ranged from 29 to 73. However, after fitting these linear regressions, we found that the models did not meet the assumption of normally distributed residuals, and so we instead fitted these models using quantile regression on the median. Model outcomes and interpretations did not meaningfully change after we switched from our preregistered analytic approach to a more appropriate modeling approach that is robust to nonnormally distributed residuals and outliers.

To test the strength of the relationship between trait and state ER diversity (exploratory Hypotheses 3a–3d), we tested the pairwise correlations between each of the trait ER diversity metrics and their parallel state diversity metric in the participants belonging to the high-social-anxiety group.

Results

Using trait ER diversity metrics to predict social-anxiety-group membership

Consistent with hypotheses, results showed that individuals who endorsed higher trait diversity within the avoidance-oriented ER strategies were significantly more likely to belong to the high-social-anxiety group relative to the low-social-anxiety group. However, neither overall ER diversity, within-approach ER diversity, nor between-factor ER diversity predicted social-anxiety-group membership when measured at a trait level. See Table 3. Given that participants in the high-social-anxiety group were significantly older than participants in the low-social-anxiety group, we added participant age as a covariate and reran these models. All results remained unchanged.

Table 3.

Results From Models Predicting Social Anxiety From Emotion Regulation (ER) Diversity Measured at Trait and State Levels

Variable	Trait ER diversity					State ER diversity
Variable	β	SE	Test statistic (Z)	p	OR	B	95% CI	B
Overall	0.56	0.37	1.53	.126	1.76	10.38	[1.27, 24.25]	0.20
Within-avoidance	1.20	0.43	2.79	.005	3.33	15.31	[4.63, 31.42]	0.25
Within-approach	−0.21	0.47	−0.44	.660	0.81	7.24	[2.49, 16.62]	0.18
Between-factor	0.02	1.72	0.01	.989	1.02	24.17	[1.39, 50.90]	0.22

Note: Trait ER diversity analyses predicted social-anxiety-group membership (high vs. low). State ER diversity analyses predicted social anxiety severity amongst participants in the high-social-anxiety group. State ER diversity results reflect the effect of the given state ER diversity score on social anxiety severity after controlling for the number of ecological momentary assessment surveys submitted by each participant. Significant results are given in boldface. Significance values are not available for quantile regressions on the median. For state analyses, ER diversity metrics are significant if the lower and upper bounds for the 95% confidence interval (CI) do not span zero. OR = odds ratio; B = non-standardized beta estimate.

Using state ER diversity metrics to predict severity of social anxiety in the high-social-anxiety group

Consistent with hypotheses and what was observed at the trait level, results showed that individuals demonstrating higher state ER diversity within the avoidance-oriented strategies reported relatively higher (vs. lower) social anxiety severity. However, neither overall ER diversity, within-approach ER diversity, nor between-factor ER diversity significantly predicted trait social anxiety severity when these diversity metrics were measured at a state level. That said, given that the number of observations that contributed to the state ER diversity scores varied between participants, we reran these analyses controlling for each participant’s number of survey observations. After controlling for number of survey responses, we found that all state ER diversity metrics significantly and positively predicted social anxiety severity (see Table 3). These updated results were mixed with respect to our hypotheses: Consistent with hypotheses, results showed that individuals who reported greater diversity in their state ER strategy choices, both overall and within the avoidance-oriented strategies, reported relatively higher (vs. lower) social anxiety severity. Inconsistent with hypotheses, however, results showed that greater state within-approach and greater state between-factor ER diversity scores were also associated with higher (vs. lower) social anxiety severity. See Table 3.

To preliminarily test the possibility that all four of these ER diversity constructs were positively associated with social anxiety severity due to co-occurring worse state affect in participants with higher social anxiety severity, we added each individual’s average state affect into the models as a centered predictor and reran these analyses. After controlling for both number of EMA surveys and average state affect, we found that the effect of within-avoidance state ER diversity on social anxiety severity no longer remained significant (b = 13.58, 95% confidence interval [CI] = [−3.39, 25.16]). However, all other state ER diversity effects remained significant in the same direction. As expected, more negative average state affect was associated with higher social anxiety severity in all models.

Strength of association between trait and state ER diversity metrics in the high-social-anxiety group

Only the correlation between trait and state within-avoidance ER diversity scores was significant (r = .27, p < .01). The correlations between trait and state overall ER diversity (r = .15, p = .127), within-approach ER diversity (r = .09, p = .372), and between-factor diversity (r = .13, p = .193) were not significant.

Secondary analyses

Given that social anxiety is highly comorbid with depression (Langer & Rodebaugh, 2014) and that ER diversity has been shown to predict depression diagnostic status (Wen et al., 2021), we ran multivariate regression analyses in lavaan (Rosseel, 2012) in which each ER diversity metric simultaneously predicted both social anxiety and depression severity. To account for the positive association between depression and social anxiety, we allowed these two variables to correlate in all models. In all models, depression severity was measured continuously with the PHQ-8. In the models using trait-based ER diversity metrics, social anxiety was measured as membership in the high-severity (1) or low-severity (0) group. In the models using state-based ER diversity metrics, social anxiety severity was measured continuously with the SIAS.

Trait ER diversity metrics

Consistent with our primary analyses, results showed that individuals who endorsed higher trait diversity within the avoidance-oriented ER strategies remained significantly more likely to belong to the high-social-anxiety group relative to the low-social-anxiety group. These individuals were also more likely to report greater depression severity. Although overall trait ER diversity continued not to differentiate between high- and low-social-anxiety group members, individuals who reported higher overall trait ER diversity were significantly more likely to report greater depression severity. Neither within-approach ER diversity nor between-factor ER diversity predicted social-anxiety-group membership or depression severity when measured at a trait level. See Table 4.

Table 4.

Results From Models Simultaneously Predicting Social Anxiety and Depression From Emotion Regulation (ER) Diversity Measured at Trait and State Levels

	Estimate	SE	Test statistic (Z)	p
Trait overall ER diversity
Social-anxiety group	b = 0.35	0.23	1.52	.130
Patient Health Questionnaire	b = 2.63	0.73	3.60	< .001
Trait within-avoidance
Social-anxiety group	b = 0.72	0.24	3.05	.002
Patient Health Questionnaire	b = 3.97	0.89	4.49	< .001
Trait within-approach
Social-anxiety group	b = −0.12	0.32	−0.38	.701
Patient Health Questionnaire	b = 0.41	0.85	0.48	.632
Trait between-factor
Social-anxiety group	b = 0.01	1.18	0.01	.992
Patient Health Questionnaire	b = 0.25	2.12	0.12	.907
State overall ER diversity
Social Interaction Anxiety Scale	β = 11.95	4.79	2.49	.013
Patient Health Questionnaire	β = 2.81	2.73	1.03	.303
State within-avoidance
Social Interaction Anxiety Scale	β = 18.88	5.65	3.34	.001
Patient Health Questionnaire	β = 6.07	3.25	1.86	.062
State within-approach
Social Interaction Anxiety Scale	β = 8.24	3.72	2.22	.027
Patient Health Questionnaire	β = 2.27	2.10	1.08	.281
State between-factor
Social Interaction Anxiety Scale	β = 23.67	10.15	2.33	.020
Patient Health Questionnaire	β = 5.67	5.76	0.99	.325

Note: Social anxiety group was coded 1 (high) or 0 (low). Social anxiety and depression were significantly positively correlated in all models. Significant results are given in boldface.

State ER diversity metrics

Consistent with our primary analyses, results showed that individuals who reported greater diversity in their state ER strategy choices—according to all four metrics of state diversity—reported higher social anxiety severity on the SIAS. Surprisingly, none of these state ER diversity metrics significantly predicted depression severity. See Table 4.

Discussion

The present study tested the association between ER diversity and social anxiety severity. We measured ER diversity, operationalized as the variety, frequency, and evenness of ER strategy use, among a sample of individuals with either high or low social anxiety. Our work built on Wen and colleagues’ (2021) mathematical operationalizations of trait-level ER diversity by developing ER diversity metrics for repeated measures data. We further expanded on Wen et al.’s (2021) work, which investigated ER diversity in depression, by examining ER diversity in social anxiety and considering diversity within and between avoidance- and approach-oriented strategies. We found that, at the trait level, higher diversity within avoidance-oriented strategies predicted belonging to the high- versus low-social-anxiety group. At the state level, higher diversity across all strategies, within avoidance- and approach-oriented strategies, and between avoidance- and approach-oriented strategy factors all predicted higher social anxiety severity among individuals in the high-social-anxiety group (but only when analyses covaried for the number of EMA surveys submitted by each individual; additionally, see nuanced results when analyses controlled for average state negative affect). Lastly, out of all state–trait ER diversity pairings, only ER diversity within avoidance-oriented strategies was significantly correlated.

Only trait within-avoidance ER diversity predicted social-anxiety-group membership

We tested whether four metrics of trait ER diversity significantly predicted an individual’s likelihood of belonging to either a high- or a low-social-anxiety group. Only diversity within avoidance-oriented strategies significantly predicted social-anxiety-group membership. Specifically, individuals who reported more evenly and frequently using a variety of avoidance-oriented strategies were significantly more likely to belong to the high- versus low-social-anxiety group. This finding is in line with our hypotheses and suggests that ER diversity within avoidance-oriented strategies may be a marker of social anxiety, especially given that this association remained significant in analyses simultaneously predicting depression severity and allowing depression and social anxiety group to correlate. Further, it fits well with prior evidence that individuals with higher SAD symptoms are avoidant regulators (Goodman et al., 2021). This finding is also in line with Wen and colleagues’ (2021) report that more diverse use of maladaptive strategies, which are typically avoidance-oriented, characterizes individuals with current and remitted depression—a finding that we conceptually replicated in our own data after including depression severity as an additional outcome in all analyses.

The three remaining trait ER diversity metrics did not significantly predict social-anxiety-group membership—although individuals who reported higher overall trait ER diversity were significantly more likely to report greater depression severity, which conceptually replicates findings from Wen and colleagues (2021). These null social anxiety results suggest that, except for avoidance-oriented strategies, individuals with higher social anxiety use a diversity of ER strategies (including approach-oriented ones) with similar frequency as do individuals with lower social anxiety. This is inconsistent with our prediction that high socially anxious individuals would be less diverse in their use of approach-oriented strategies. However, there is prior evidence that individuals higher in SAD symptoms use reappraisal (a cognitive approach-oriented strategy) at similar rates as do individuals lower in SAD symptoms (see Dryman & Heimberg, 2018). Thus, it is possible that the frequency (and therefore the diversity) with which individuals use a range of approach-oriented strategies does not differentiate between high- and low-SAD groups; perhaps, instead, differences in the skillful selection and use of approach-oriented strategies varies (Dryman & Heimberg, 2018; Farmer & Kashdan, 2012). Further, we assessed avoidance-oriented strategies that were relatively more behavioral (e.g., eating, sleeping, doing something fun with others) than the approach-oriented strategies (e.g., acceptance, reappraisal). Given that people tend to recall and report behavioral activities more accurately than cognitive activities (Nisbett & Wilson, 1977; Vazire & Mehl, 2008), data on the behavioral, avoidance-oriented strategies may have allowed for a more reliable comparison between the two groups.

All state ER diversity metrics predicted social anxiety severity in the high-social-anxiety group

Prior to controlling for the number of EMA surveys submitted by each participant, we found that only state-level within-avoidance ER diversity significantly predicted social anxiety severity. Interestingly, although the effect of how many EMA surveys were submitted was not itself a significant predictor of social anxiety severity, accounting for the variation explained by this EMA survey count variable allowed all four state-level ER diversity variables to significantly predict the residual variance in social anxiety severity. Specifically, we found that all state-level ER diversity metrics predicted higher levels of social anxiety in the high-social-anxiety group after we covaried the number of EMA surveys submitted by the individual. Two of these findings are consistent with our hypotheses. First, we found that higher state within-avoidance ER diversity predicted higher social anxiety severity. This finding mirrors the pattern observed at the trait level and in prior research (Goodman et al., 2021), indicating that participants higher in social anxiety severity both tend to see themselves as using a variety of avoidance-oriented strategies more frequently and evenly, in general (i.e., at a trait level), and report using these strategies more frequently and evenly in response to daily emotions. It is also in line with literature on ER profiles—another approach to capturing the relative frequency, though not the evenness, of individuals’ ER strategy use—which finds that individuals who typically use avoidance-oriented strategies report greater anxiety symptoms (Chesney et al., 2019).

Second, we found that higher overall state ER diversity predicted higher social anxiety severity. This supports one of our competing hypotheses (2b) and suggests that individuals higher (vs. lower) in social anxiety pull from a greater number of strategies and use them more evenly in response to distressing situations. This is consistent with socially anxious people using more regulatory strategies per stressful event than healthy control participants (Goodman et al., 2021). It is also consistent with work on ER profiles, which indicates that individuals who endorse frequently using multiple ER strategies report higher anxiety (Dixon-Gordon et al., 2015). However, it is not consistent with other ER profiles work citing positive emotional consequences tied to using many ER strategies (e.g., Eftekhari et al., 2009; Grommisch et al., 2020) or findings suggesting that greater between-strategy variability (Blanke et al., 2020) or reporting more unique transitions between many ER strategies (Daniel et al., 2022) is associated with lower negative affect levels. These mixed findings suggest that the skill with which strategies are used, or fit of the strategy to the situation, may be more important considerations for determining adaptiveness of ER than variability alone.

The two remaining findings are inconsistent with our hypotheses. We found that higher state ER diversity within approach-oriented ER strategies and between factors predicted higher social anxiety severity. This suggests that individuals with higher social anxiety severity reported greater variety and more even, frequent use of approach-oriented strategies, as well as greater variety and more even, frequent use of both avoidance- and approach-oriented strategies. These findings do not align with ER flexibility conceptualizations (Wen et al., 2021) that higher ER diversity in approach-oriented strategies and between approach- and avoidance-oriented strategies are markers of better mental health outcomes. One plausible explanation for the current findings is that socially anxious individuals in our sample experienced more anxiety at the state level and therefore called on a greater number of strategies overall. This seems likely, given evidence that individuals with SAD endorse higher rates of state anxiety (Kashdan et al., 2014) and typically report difficulty tolerating negative emotions (Keough et al., 2010). Motivation to use any possible strategy to try to make anxiety and negative emotions go away as fast as possible would result in greater endorsements of strategies within and across strategy factors.

This interpretation raises an interesting possibility that individuals with higher social anxiety may draw on approach-oriented strategies and use them to minimize contact with negative emotions as quickly as possible (e.g., similar to using worry to avoid intense negative emotion; Borkovec et al., 2004). This would fit well with recent work by Tifft and colleagues (2022), who found that more than half of their participants engaged in meditation with control/avoidance-based (vs. acceptance-based) intentions and that they experienced greater worry, anxiety, and negative affect compared with those who engaged in meditation with acceptance-based intentions. Additionally, in our sample, including average state negative affect as a simultaneous predictor made within-avoidance state-level ER diversity no longer significant, suggesting that participants’ negative affect levels at least partly motivated their choice of (avoidance-oriented) ER strategies. However, the current findings also indicate that although participants with higher social anxiety severity did report higher average state negative affect throughout the EMA, including average state negative affect as a simultaneous predictor did not diminish the significant effects of the overall, within-approach, and between avoidance and approach state-level ER diversity metrics in predicting higher social anxiety severity. Further research is therefore needed to understand how highly socially anxious individuals vary their approach- and avoidance-oriented strategies and what they intend to accomplish when using different strategies.

Trait and state ER diversity metrics were largely unrelated

Our exploratory analyses found that only the correlation between trait and state within-avoidance ER diversity scores was significant, and it was small to moderate in size. This finding fits with prior research reporting modest associations between trait and state ER reports (Blalock et al., 2016; Brockman et al., 2017). Finding greater concordance between trait and state avoidance-oriented ER diversity (vs. other ER diversity metrics) further suggests that our sample may have found it easier to accurately introspect and report on their use of avoidance strategies. Similarly, it is possible that highly anxious individuals’ perceptions of how they typically regulate (i.e., their trait ER diversity) underestimate their approach-oriented ER efforts because these involve fewer concrete behavioral markers and may be more likely to occur during less saliently intense moments of emotional arousal. If recall or reporting biases led to underreporting of trait approach-oriented ER strategy use, then this may explain the reduced association between state and trait approach-oriented ER diversity metrics.

Clinical implications

The present findings have at least three implications for treatment. First, they suggest that individuals with high social anxiety severity use a variety of both approach- and avoidance-oriented strategies in daily life rather than exclusively avoidant strategies. This suggests that individuals with relatively high social anxiety scores may not have trouble deploying a variety of strategies. However, ER diversity does not consider skillfulness of ER strategy use or ability to deploy the “right” strategy to fit the ongoing demands of the situation. This may point to a need for interventions that target skill building in strategy use or context-strategy fit rather than willingness or even tendency to implement many different strategies.

Second, our findings raise interesting questions regarding the association of ER diversity and social anxiety within different time frames. Whereas at the trait-level, only avoidance-oriented ER diversity was positively associated with the likelihood of belonging to the high-social-anxiety group, all four diversity constructs assessed at the state level in daily life were positively associated with social anxiety severity (when analyses controlled for number of EMA surveys). Although methodological differences could account for this different pattern of findings (i.e., low/high group comparison vs. dimensional comparison within the high-social-anxiety group), if replicated, this might point to the importance of considering time frame when conceptualizing the role of ER diversity. For example, showing higher between-factor diversity over longer periods of time (e.g., across different days or weeks) may be associated with positive mental health outcomes by capturing an ability to match ER strategy use to varied situational demands. Meanwhile, showing higher between-factor diversity over shorter periods of time (e.g., within the same hour) may be associated with negative mental health outcomes by capturing a tendency to “flail” and not adequately persevere on strategy implementation (Southward et al., 2018). Given these possible interpretations of high ER diversity, clinicians assessing this construct may find it useful to consider the measurement time frame (e.g., trait vs. daily reporting) when incorporating this information into treatment planning. For example, clinicians might probe their client’s ER strategy selection process if they notice their client report high ER diversity within a short time frame, as this might suggest that their client is flailing in their ER strategy selections more so than if their client had reported high ER diversity across a longer time frame.

Third, our secondary analyses found that higher trait within-avoidance ER diversity simultaneously predicted both higher social anxiety and higher depression severity, which suggests that both higher social anxiety and higher depression are each associated with greater reliance on a diversity of avoidance-oriented strategies. This is useful information for clinicians, especially given high comorbidity among these symptom groups (Langer & Rodebaugh, 2014). However, findings also indicate that ER diversity does not operate identically within social anxiety and depression. For example, trait overall ER diversity was positively associated with depression but not with social anxiety severity. Further, although the state ER diversity metrics were each significantly associated with social anxiety, none were associated with depression. This suggests that treatment may need to be approached differently when targeting ER difficulties between these disorders (e.g., targeting depressive withdrawal vs. anxious avoidance).

Limitations and future directions

We encourage readers to interpret the present findings considering several limitations. First, our operationalization of state-level ER diversity does not account for how strategically matched the ER strategies used are to the situation at hand. Thus, these ER diversity metrics provide insight into whether individuals exhibit the ER variation we would expect of a healthy regulator but not whether they are optimally matching their ER strategies to their contexts.

Second, we conceptually categorized strategies as approach versus avoidance, but we did not directly assess the actual approach/avoidance impact of each strategy. Although our classification approach was based on empirical support (e.g., Daros et al., 2019; McMahon & Naragon-Gainey, 2019) and the relevance of emotional avoidance to SAD (American Psychiatric Association, 2013), we cannot know whether participants’ use of a given strategy did indeed lead them to be engaged with their emotion or distanced from their emotion. Given that the way strategies are classified influences the association between ER diversity and mental health constructs (Wen et al., 2021), future research may benefit from assessing participants’ ER goals and intentions (Eldesouky & Gross, 2019; Tifft et al., 2022) as well as actual approach/avoidance impact to further explore optimal ER classification systems. Efforts to improve this classification system might also benefit from testing whether the specific emotion/affect valence being regulated (e.g., anxiety vs. sadness) changes the classification. Relatedly, we assessed frequency of ER strategy use with measures that used only one item to assess each strategy because this allowed us to maximize sampling a broad range of strategies at both trait and state levels while minimizing user burden on the EMA surveys. Although this limits the opportunity for tests of validity and reliability, we used face-valid items that we pilot-tested with a team of research assistants and that mirrored language from previous EMA studies assessing the use of many ER strategies (e.g., Heiy & Cheavens, 2014).

Third, because of our study design, we were unable to calculate state ER diversity metrics among individuals with low social anxiety severity. By examining these associations only among individuals with high social anxiety, we may have suppressed the relationships between state ER diversity and social anxiety severity; however, the range of SIAS scores in the high-social-anxiety group was still broad (i.e., 29–73 out of 0–80).

Fourth, our ER diversity metrics did not consider individuals’ state anxiety specific to when they regulated their emotions or the actual skill with which they used the ER strategies they reported. These will be important extensions for future work given research suggesting that emotional intensity in part guides ER strategy selection (e.g., O’Toole et al., 2017) and that individuals with psychopathology may use strategies frequently but ineffectively (e.g., Kivity & Huppert, 2016). The state equations we introduced offer useful starting points for these extensions. For example, researchers might incorporate state affect by conditioning each probability on the expected value of state affect at that time. However, doing this effectively would require researchers to find a principled way to determine the expected value of affect by deciding whether they will allow the expected value to vary over time, to vary between ER strategies, and to vary between people, among other considerations. Given the complexity of these extensions, we elected to offer our foundational approach first.

Fifth, although we recruited individuals with high and low social anxiety symptoms, we did not have a clinical sample. Therefore, our study was not positioned to examine whether ER diversity is a marker of SAD diagnostic status. Relatedly, our sample was composed of primarily female, non-Hispanic White, young, highly educated individuals. Therefore, our findings may not generalize to individuals who hold different identities. Additionally, although all participants were living in the Southeastern United States during their participation in this study, we did not gather data regarding participants’ cultural backgrounds, which further limits the generalizability of these findings. Future research is needed in clinical and more diverse, representative samples.

Finally, although previous analyses in these data have found that participants were more likely to report regulating their emotions when also reporting relatively higher state negative affect (Daniel, Goodman, et al., 2020), the survey’s design left open the possibility that participants could also report regulation attempts intended to manage their positive emotions. Thus, regulation of positive and negative affect are treated equivalently in the current work. Given that social anxiety has been associated with deficits in maintaining positive affect (Farmer et al., 2014), future work that can disentangle the relationship between ER diversity when regulating positive versus negative affect might provide greater understanding of ER differences in social anxiety. For example, it is possible that socially anxious people have limited diversity in strategies to upregulate positive emotions, potentially contributing to positive emotion maintenance deficits, but greater diversity in strategies to downregulate negative emotions.

Conclusion

In the present study, we extended mathematical operationalizations of ER diversity (Wen et al., 2021) for use with intensive longitudinal data and tested how trait- and state-level ER diversity predict social anxiety severity. Across trait and state levels, higher diversity of avoidance-oriented strategies predicted higher social anxiety severity, suggesting that higher ER diversity within avoidance-oriented strategies may be a marker of SAD. At the state level, higher ER diversity across all strategies, within approach-oriented strategies, and between approach- and avoidance-oriented strategy groups also predicted higher social anxiety severity among a high-social-anxiety-symptoms group, but only after analyses accounted for the number of EMA surveys submitted by each participant. This suggests that individuals with greater social anxiety severity may frequently resort to a variety of strategies—both avoidance- and approach-oriented—to reduce state distress, even if they do not report this pattern when asked about their ER tendencies in general. These findings raise interesting questions regarding when ER diversity is a marker of emotional health versus dysregulation. Along these lines, future work may benefit from assessing the strategic selection of ER strategies (e.g., match to situational demands) and their skillful implementation to learn more about when choosing a diversity of strategies reflects healthy flexibility versus desperate scrambling to escape negative affect.

Footnotes

Acknowledgements

We thank Miranda L. Beltzer and Alexander R. Daros for their help collecting the data for this study and Robert G. Moulder for his help building the ER diversity R functions used to preprocess these data.

Transparency

Action Editor: Jennifer Lau

Editor: Jennifer L. Tackett

Author Contributions

Katharine E. Daniel: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Project administration; Writing – original draft.

Maria A. Larrazabal: Conceptualization; writing – original draft.

Mehdi Boukhechba: Investigation; Supervision; Writing – review & editing.

Laura E. Barnes: Funding acquisition; Investigation; Supervision; Writing – review & editing.

Bethany A. Teachman: Conceptualization; Funding acquisition; Investigation; Supervision; Writing – original draft.

ORCID iDs

Katharine E. Daniel

Laura E. Barnes

Notes

References

Aldao

Dixon-Gordon

K. L.

(2014). Broadening the scope of research on emotion regulation strategies and psychopathology. Cognitive Behavior Therapy, 43(1), 22–33. https://doi.org/10.1080/16506073.2013.816769

Aldao

Sheppes

Gross

J. J.

(2015). Emotion regulation flexibility. Cognitive Therapy and Research, 39, 263–278. https://doi.org/10.1007/s10608-014-9662-4

American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (5th ed.). https://doi.org/10.1176/appi.books.9780890425596

Beltzer

M. L.

(2022). Examining social reinforcement learning biases in social anxiety [Doctoral dissertation, University of Virginia]. Libra. https://doi.org/10.18130/gfck-0g65

Blalock

D. V.

Kashdan

T. B.

Farmer

A. S.

(2016). Trait and daily emotion regulation in social anxiety disorder. Cognitive Therapy and Research, 40, 416–425. https://doi.org/10.1007/s10608-015-9739-8

Blanke

E. S.

Brose

Kalokerinos

E. K.

Erbas

Riediger

Kuppens

(2020). Mix it to fix it: Emotion regulation variability in daily life. Emotion, 20(3), 473–485. https://doi.org/10.1037/emo0000566

Bonanno

G. A.

Burton

C. L.

(2013). Regulatory flexibility: An individual differences perspective on coping and emotion regulation. Perspectives on Psychological Science, 8(6), 591–612. https://doi.org/10.1177/1745691613504116

Borkovec

T. D.

Alcaine

Behar

(2004). Avoidance theory of worry and generalized anxiety disorder. In Heimberg

R. G.

Turk

C. L.

Mennin

D. S.

(Eds.), Generalized anxiety disorder: Advances in research and practice (pp. 77–108). Guilford Press.

Brockman

Ciarrochi

Parker

Kashdan

(2017). Emotion regulation strategies in daily life: Mindfulness, cognitive reappraisal and emotion suppression. Cognitive Behaviour Therapy, 46(2), 91–113. https://doi.org/10.1080/16506073.2016.1218926

10.

Chesney

S. A.

Timmer-Murillo

S. C.

Gordon

N. S.

(2019). Establishment and replication of emotion regulation profiles: Implications for psychological health. Anxiety, Stress, & Coping, 32(3), 329–345.

11.

Daniel

K. E.

Daros

A. R.

Beltzer

M. L.

Boukhechba

Barnes

L. E.

Teachman

B. A.

(2020). How anxious are you right now? Using ecological momentary assessment to evaluate the effects of cognitive bias modification for social threat interpretations. Cognitive Therapy and Research, 44(3), 538–556. https://doi.org/10.1007/s10608-020-10088-2

12.

Daniel

K. E.

Goodman

F. R.

Beltzer

M. L.

Daros

A. R.

Boukhechba

Barnes

L. E.

Teachman

B. A.

(2020). Emotion malleability beliefs and emotion experience and regulation in the daily lives of people with high trait social anxiety. Cognitive Therapy and Research, 44, 1186–1198. https://doi.org/10.1007/s10608-020-10139-8

13.

Daniel

K. E.

Moulder

R. G.

Teachman

B. A.

Boker

S. M.

(2022). Stability and spread: A novel method for quantifying transitions within multivariate binary time series data. Behavior Research Methods. Advance online publication. https://doi.org/10.3758/s13428-022-01942-0

14.

Daros

A. R.

Daniel

K. E.

Meyer

M. J.

Chow

P. I.

Barnes

L. E.

Teachman

B. A.

(2019). Impact of social anxiety and social context on college students’ emotion regulation strategy use: An experience sampling study. Motivation and Emotion, 43, 844–855. https://doi.org/10.1007/s11031-019-09773-x

15.

Dixon-Gordon

K. L.

Aldao

De Los Reyes

(2015). Repertoires of emotion regulation: A person-centered approach to assessing emotion regulation strategies and links to psychopathology. Cognition and Emotion, 29(7), 1314–1325. https://doi.org/10.1080/02699931.2014.983046

16.

Dryman

M. T.

Heimberg

R. G.

(2018). Emotion regulation in social anxiety and depression: A systematic review of expressive suppression and cognitive reappraisal. Clinical Psychology Review, 65, 17–42. https://doi.org/10.1016/j.cpr.2018.07.004

17.

Eftekhari

Zoellner

L. A.

Vigil

S. A.

(2009). Patterns of emotion regulation and psychopathology. Anxiety, Stress, & Coping, 22(5), 571–586. https://doi.org/10.1080/10615800802179860

18.

Eldesouky

Gross

J. J.

(2019). Emotion regulation goals: An individual difference perspective. Social and Personality Psychology Compass, 13(9), Article e12493. https://doi.org/10.1111/spc3.12493

19.

Farmer

A. S.

Kashdan

T. B.

(2012). Social anxiety and emotion regulation in daily life: Spillover effects on positive and negative social events. Cognitive Behaviour Therapy, 41(2), 152–162. https://doi.org/10.1080/16506073.2012.666561

20.

Farmer

A. S.

Kashdan

T. B.

Weeks

J. W.

(2014). Positivity deficits in social anxiety: Emotions, events, and cognitions. In Hofmann

S. G.

DiBartolo

P. M.

(Eds.), Social anxiety (3rd ed.; pp. 551–578). Academic Press. https://doi.org/10.1016/B978-0-12-394427-6.00019-4

21.

Goodman

F. R.

Daniel

K. E.

Eldesouky

Brown

B. A.

Kneeland

E. T.

(2021). How do people with social anxiety disorder manage daily stressors? Deconstructing emotion regulation flexibility in daily life. Journal of Affective Disorders Reports, 6, Article 100210. https://doi.org/10.1016/j.jadr.2021.100210

22.

Goodman

F. R.

Kashdan

T. B.

(2021). The most important life goals of people with and without social anxiety disorder: Focusing on emotional interference and uncovering meaning in life. The Journal of Positive Psychology, 16(2), 272–281. https://doi.org/0.1080/17439760.2019.1689423

23.

Grommisch

Koval

Hinton

J. D. X.

Gleeson

Hollenstein

Kuppens

Lischetzke

(2020). Modeling individual differences in emotion regulation repertoire in daily life with multilevel latent profile analysis. Emotion, 20(8), 1462–1474. https://doi.org/10.1037/emo0000669

24.

Heiy

J. E.

Cheavens

J. S.

(2014). Back to basics: A naturalistic assessment of the experience and regulation of emotion. Emotion, 14(5), 878–891. https://doi.org/10.1037/a0037231

25.

Jazaieri

Morrison

A. S.

Goldin

P. R.

Gross

J. J.

(2015). The role of emotion and emotion regulation in social anxiety disorder. Current Psychiatry Reports, 17, Article 531. https://doi.org/10.1007/s11920-014-0531-3

26.

Kashdan

T. B.

Farmer

A. S.

Adams

L. M.

Ferssizidis

McKnight

P. E.

Nezlek

J. B.

(2013). Distinguishing healthy adults from people with social anxiety disorder: Evidence for the value of experiential avoidance and positive emotions in everyday social interactions. Journal of Abnormal Psychology, 122(3), 645–655. https://doi.org/10.1037/a0032733

27.

Kashdan

T. B.

Goodman

F. R.

Machell

K. A.

Kleiman

E. M.

Monfort

S. S.

Ciarrochi

Nezlek

J. B.

(2014). A contextual approach to experiential avoidance and social anxiety: Evidence from an experimental interaction and daily interactions of people with social anxiety disorder. Emotion, 14(4), 769–781. https://doi.org/10.1037/a0035935

28.

Kashdan

T. B.

Weeks

J. W.

Savostyanova

A. A.

(2011). Whether, how, and when social anxiety shapes positive experiences and events: A self-regulatory framework and treatment implications. Clinical Psychology Review, 31(5), 786–799. https://doi.org/10.1016/j.cpr.2011.03.012

29.

Keough

M. E.

Riccardi

C. J.

Timpano

K. R.

Mitchell

M. A.

Schmidt

N. B.

(2010). Anxiety symptomatology: The association with distress tolerance and anxiety sensitivity. Behavior Therapy, 41, 567–574.

30.

Kivity

Huppert

J. D.

(2016). Does cognitive reappraisal reduce anxiety? A daily diary study of a micro-intervention with individuals with high social anxiety. Journal of Consulting and Clinical Psychology, 84(3), 269–283. https://doi.org/10.1037/ccp0000075

31.

Kroenke

Strine

T. W.

Spitzer

R. L.

Williams

J. B.

Berry

J. T.

Mokdad

A. H.

(2009). The PHQ-8 as a measure of current depression in the general population. Journal of Affective Disorders, 114(1–3), 163–173. https://doi.org/10.1016/j.jad.2008.06.026

32.

Langer

J. K.

Rodebaugh

T. L.

(2014). Comorbidity of social anxiety disorder and depression. In Richards

C. S.

O’Hara

M. W.

(Eds.), The Oxford handbook of depression and comorbidity (pp. 111–128). Oxford University Press.

33.

Mattick

R. P.

Clarke

J. C.

(1998). Development and validation of measures of social phobia scrutiny fear and social interaction anxiety. Behaviour Research and Therapy, 36(4), 455–470. https://doi.org/10.1016/s0005-7967(97)10031-6

34.

McMahon

T. P.

Naragon-Gainey

(2019). The multilevel structure of daily emotion-regulation-strategy use: An examination of within- and between-person associations in naturalistic settings. Clinical Psychological Science, 7(2), 321–339. https://doi.org/10.1177/2167702618807408

35.

Nisbett

R. E.

Wilson

T. D.

(1977). Telling more than we can know: Verbal reports on mental processes. Psychological Review, 84(3), 231–259.

36.

O’Toole

M. S.

Zachariae

Mennin

D. S.

(2017). Social anxiety and emotion regulation flexibility: Considering emotion intensity and type as contextual factors. Anxiety, Stress, & Coping, 30, 716–724. https://doi.org/10.1080/10615806.2017.1346792

37.

Rapee

R. M.

Heimberg

R. G.

(1997). A cognitive-behavioral model of anxiety in social phobia. Behaviour Research and Therapy, 35(8), 741–756. https://doi.org/10.1016/s0005-7967(97)00022-3

38.

R Core Team. (2017). R: A language and environment for statistical computing [Computer software]. https://www.R-project.org

39.

Rosseel

(2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2). https://doi.org/10.18637/jss.v048.i02

40.

Rusch

Westermann

Lincoln

T. M.

(2012). Specificity of emotion regulation deficits in social anxiety: An Internet study. Psychology and Psychotherapy: Theory, Research and Practice, 85(3), 268–277. https://doi.org/10.1111/j.2044-8341.2011.02029.x

41.

Southward

M. W.

Altenburger

E. M.

Moss

S. A.

Cregg

D. R.

Cheavens

J. S.

(2018). Flexible, yet firm: A model of healthy emotion regulation. Journal of Social and Clinical Psychology, 37(4), 231–251. https://doi.org/10.1521/jscp.2018.37.4.231

42.

Tifft

E. D.

Underwood

S. B.

Roberts

M. Z.

Forsyth

J. P.

(2022). Using meditation in control vs. acceptance context: A preliminary evaluation of relations with anxiety, depression, and indices of well-being. Journal of Clinical Psychology, 78(7), 1407–1421. https://doi.org/10.1002/jclp.23313

43.

Turk

C. L.

Heimberg

R. G.

Luterek

J. A.

Mennin

D. S.

Fresco

D. M.

(2005). Emotion dysregulation in generalized anxiety disorder: A comparison with social anxiety disorder. Cognitive Therapy and Research, 29, 89–106. https://doi.org/10.1007/s10608-005-1651-1

44.

Vazire

Mehl

M. R.

(2008). Knowing me, knowing you: The accuracy and unique predictive validity of self-ratings and other-ratings of daily behavior. Journal of Personality and Social Psychology, 95(5), 1202–1216. https://doi.org/10.1037/a0013314

45.

Wen

Quigley

Yoon

K. L.

Dobson

K. S.

(2021). Emotion regulation diversity in current and remitted depression. Clinical Psychological Science, 9(4), 563–578. https://doi.org/10.1177/2167702620978616