The Multilevel Structure of Daily Emotion-Regulation-Strategy Use: An Examination of Within- and Between-Person Associations in Naturalistic Settings

Abstract

Existing structural analyses of emotion-regulation (ER) strategies have relied on retrospective, dispositional assessments, ignoring the within-person structure (i.e., intraindividual strategy groupings based on momentary covariances) and variability in strategy use across time and contexts. We conducted multilevel exploratory factor analyses on self-reported daily use of 11 strategies (i.e., acceptance, behavioral avoidance, distraction, experiential avoidance, expressive suppression, procrastination, reappraisal, reflection, rumination, savoring, social support) in clinical (N = 129) and student (N = 109) samples with intensive longitudinal designs. At the between-person level, two factors—Engagement and Avoidance—emerged in both samples. A different structure was found at the within-person level, with four factors in the student sample (i.e., Attentional Shift, Acceptance, Avoidance, Emotional Expression) and three in the clinical sample (i.e., Attentional Shift, Avoidance, Emotional Expression). The validity of these factors was examined via their associations with daily internalizing symptoms and affect. Implications for naturalistic ER strategy use and clinical assessment/intervention are discussed.

Keywords

emotion regulation strategies internalizing symptoms positive affect negative affect intensive longitudinal design open data

Emotion regulation (ER) has been described as an important transdiagnostic process relevant to the etiology of many psychological disorders and is proposed to have a mechanistic role in the effectiveness of psychotherapy (e.g., Gallagher, 2017; Moyal, Cohen, Henik, & Anholt, 2015; Sheppes, Suri, & Gross, 2015). In its broadest sense, ER can be defined as an attempt to alter the course of an emotion, including modifying or maintaining its duration, intensity, or valence (Aldao, Nolen-Hoeksema, & Schweizer, 2010; Gross, 1998, 2015; Gross & Jazaieri, 2014). Although this definition suggests that any action that aims to change how an emotional process unfolds is an ER effort (e.g., Berking & Wupperman, 2012), one branch of research focuses on specific commonly used ER strategies, or cognitive and behavioral approaches that are typically implemented with the intent of modifying one’s emotional experience. There are many different possible ER strategies (likely in the hundreds; e.g., Parkinson & Totterdell, 1999), but well-studied examples include distraction, behavioral and experiential avoidance, acceptance, rumination, mindfulness, problem solving, worry, reappraisal, and expressive suppression (e.g., Aldao et al., 2010; Gross, 2015; Naragon-Gainey, McMahon, & Chacko, 2017; Webb, Miles, & Sheeran, 2012).

Models of ER Strategies

Given the importance of ER strategies in relation to healthy psychological functioning, a number of theoretical frameworks and structural models have been proposed to consolidate research on the many individual ER strategies and characterize their role in ER more generally (see Naragon-Gainey et al., 2017, for a more detailed review). Most studies focus on one or a few strategies without reference to how they fit into the larger universe of ER strategies, so structural frameworks are necessary for connecting local literatures on specific strategies by articulating meaningful underlying groupings. For example, groups of strategies may reflect important conceptual distinctions (e.g., temporal processes, function of strategy use, formal features; Gross, 1998, 2015; Koole, 2009; Parkinson & Totterdell, 1999), empirical associations with other constructs or outcomes (e.g., associations with psychopathology; Aldao et al., 2010), or observed empirical covariances among ER strategies (Aldao & Nolen-Hoeksema, 2010; Lee, Witte, Weathers, & Davis, 2015; Naragon-Gainey et al., 2017; Seligowski & Orcutt, 2015). In particular, the latter approach—analyzing strategy covariances to reveal groups of ER strategies that tend to be used together—is important for synthesizing and refining research investigating how individuals select and use ER strategies as well as the consequences of strategy usage for well-being.

We focus our review on models that are amenable to testing via self-reported endorsement of commonly used ER strategies because this is most relevant to the current study. The cognitive-behavioral framework is one example of a conceptual/formal model, which distinguishes strategies that are more behaviorally focused or overt (e.g., behavioral avoidance, expressive suppression, actions related to problem solving, seeking social support) from those that are cognitively focused or covert (e.g., reappraisal, rumination, worry, mindfulness; e.g., Aldao & Dixon-Gordon, 2014; Parkinson & Totterdell, 1999). Gross’s (1998; revised and expanded in 2015) process model of ER, which has generated a very large body of research, focuses on how different strategies are situated within the temporal trajectory of an emotional experience. According to this model, there are five stages of ER, with each stage allowing for different ways one might respond to one’s emotions: situation selection (e.g., behavioral avoidance), situation modification (e.g., problem solving, social support), attentional deployment (e.g., experiential avoidance, rumination, worry, mindfulness, distraction), cognitive change (e.g., reappraisal, acceptance), and response modulation (e.g., expressive suppression). Finally, one of the primary frameworks for ER strategy use generated from empirical associations with outcomes is the adaptive-maladaptive strategy model. This framework labels ER strategies that have positive empirical associations with psychopathology (e.g., rumination, worry, behavioral avoidance) as maladaptive, whereas ER strategies with negative associations with psychopathology (e.g., reappraisal, problem solving, acceptance) are considered adaptive strategies (see Aldao et al., 2010, for a meta-analytic summary).

Although the aforementioned models imply, to varying degrees, expected strategy groupings based on strategy covariances with one another, few studies have directly examined the covariances of strategy use/endorsement, and most are limited by a small number of included strategies and/or shared measure variance that may mask the underlying structure (Aldao & Nolen-Hoeksema, 2010; Lee et al., 2015; Seligowski & Orcutt, 2015). Naragon-Gainey and colleagues (2017) conducted a factor analysis of 10 common ER strategies (assessed with retrospective, dispositional measures), based on meta-analytic estimates of their correlations in over 300 samples. Their aim was to examine whether and how ER strategy use could be reduced to underlying dimensions or factors that account for their covariances. Results supported a three-factor structure: Disengagement (e.g., distraction, behavioral avoidance, low mindfulness), Aversive Cognitive Perseveration (e.g., rumination, worry, experiential avoidance, low acceptance, low distraction), and Adaptive Engagement (e.g., problem solving, mindfulness, reappraisal). Although empirical associations with psychopathology were not tested in this study, these factors conceptually align somewhat with the maladaptive/adaptive framework reviewed earlier. Specifically, Aversive Cognitive Perseveration largely consisted of putatively maladaptive strategies, whereas the Adaptive Engagement factor largely consisted of putatively adaptive strategies; Disengagement did not clearly fit into either.

ER as a Dynamic and Contextualized Process

The role of context

While ER strategy use typically has been assessed as a dispositional variable that is assumed to be stable over time (similar to personality traits), recent research emphasizes the contextual and dynamic nature of how one responds to emotions (e.g., Aldao, 2013; Aldao & Nolen-Hoeksema, 2012; Aldao, Sheppes, & Gross, 2015; Bonanno & Burton, 2013). That is, rather than focusing solely on which ER strategies are typically implemented by an individual, it is also important to consider how context influences strategy use and subsequent outcomes. The effectiveness of ER strategies is affected by a variety of contextual factors, including the intensity of the emotional experience, specific ER goals, and situational demands (e.g., Aldao, 2013; Dixon-Gordon, Aldao, & De Los Reyes, 2015; English, Lee, John, & Gross, 2017; Kalokerinos, Tamir, & Kuppens, 2017; Sheppes, Scheibe, Suri, & Gross, 2011). Thus, researchers have argued that it is not the habitual use of a given strategy that is most important for understanding relevant outcomes; rather, it is the flexibility of strategy use—whereby occasional variability maps onto the variability of contextual demands—and the fit of the strategy with one’s goals, situation, and abilities that may be crucial (e.g., Aldao, 2013; Bonanno & Burton, 2013).

Intensive longitudinal studies in naturalistic settings

Whereas trait measures are unable to capture variability of strategy use (and therefore unable to examine the contexts under which a given strategy is effective or ineffective), studies using intensive longitudinal designs can provide insight into how people naturalistically select and implement ER strategies. Such studies of ER strategy use in daily life reveal that people use different strategies in different situations and at different times. In fact, intraclass correlations (ICC), which partition observed variance into within-person (i.e., differences in use within an individual across occasions) versus between-person (i.e., differences in use across people) variance, suggest that ER strategies vary substantially across occasions for a given individual (ICCs = 0.21–0.63, meaning that 37% to 79% of the observed variance is due to intraindividual variability; Brans, Koval, Verduyn, Lim, & Kuppens, 2013; Brockman, Ciarrochi, Parker, & Kashdan, 2017; Haines et al., 2016). Moreover, several intensive longitudinal studies have shown that individuals report using multiple ER strategies per occasion (e.g., Brans et al., 2013; Brockman et al., 2017; Haines et al., 2016; Heiy & Cheavens, 2014). Thus, patterns in the momentary covariance between ER strategies may reflect underlying groupings that are related to the contextual demands of use.

Examining the structure of ER strategy use repeatedly as it occurs naturalistically in daily life has several important advantages. First, such reports are likely to be more ecologically valid as they involve minimal retrospection and capture variability in strategy use over time and contexts rather than requiring an abstracted answer that summarizes over the course of one’s life (e.g., Bolger & Laurenceau, 2013; Mehl & Conner, 2012). These occasion-specific responses can be aggregated and adjusted for within-person variability to identify individual differences in strategy use, similar to global retrospective measures, but are likely to yield more reliable and valid structural analyses because they should be based on more accurate assessments of use. Perhaps even more importantly, an intensive longitudinal design provides a better understanding of how individuals choose to regulate their emotions differently across different situations. Whereas trait or dispositional measures of ER strategy use can model the covariances of strategy use across individuals (i.e., individuals who tend to employ one strategy also tend to employ other strategies), intensive longitudinal designs allow for the analysis of within-person variability and how strategies covary across contexts (i.e., when a person implements one strategy, he or she also tends to implement other strategies on the same occasion). This within-person structure more closely captures causal processes in understanding how individuals change over time. Delineating both individual differences and intrapersonal variability in the use of ER strategies is essential for studying why and how individual strategies are related to one another and a broad range of outcomes.

Although the momentary, real-life covariance of ER strategies has not been studied (to our knowledge), Roesch and colleagues (2010) examined daily reports of coping strategies, a construct similar to ER,¹ using a daily diary design in a student sample. Their daily coping strategies were linked to stressors specifically (rather than to a broader range of emotion-eliciting stimuli), and they included ER strategies such as avoidance, reappraisal, problem solving, social support, and rumination, but also other coping-related items such as humor, religious coping, and physical release of emotions. Multilevel factor analysis was used, which allows for the independent examination of within-person and between-person structures such that they need not have the same number of factors or content. In other words, the covariances or groupings of strategies may be different when focusing on how an individual uses different strategies over time (within-person), as opposed to how different people tend to use different strategies (between-person). Roesch and colleagues found evidence for a four within- and four between-person factor structure. At the within-person level, the factors were interpreted as Minimization of Stressors, Problem-Focused Coping, Seeking Social Support, and Emotional Rumination. At the between-person level, two factors were similar, but two were distinct (factors were labeled Avoidance Coping, Problem-Focused Coping, Seeking Social Support, and Distraction), indicating different structures for within-person use than between-person use. This study provides an important initial examination of the intensive longitudinal structure of daily coping strategies, but it has not yet been replicated, and coping was assessed at few time points (i.e., five), limiting power and generalizability. In addition, they used a nonclinical sample, and patterns of strategy use may differ in a more distressed clinical sample.

The Current Study

The purpose of the present investigation is to examine the structure of ER strategy use in daily life to better understand how specific strategies co-occur—both with regard to individual differences in strategy use and variability in use within an individual over time. There are many theories and much interest in how, when, and why ER strategies are enacted, as well the implications of these behaviors for psychopathology, well-being, social relationships, physical health, and the trajectory of an emotional experience (e.g., Gross, 1998, 2015; Gross & Jazaieri, 2014). But without a basic understanding of which strategies tend to be enacted together in daily life, it is difficult to systematically study these questions, particularly given the multitude of possible strategies. The current study aimed to provide such an initial framework for future research. In addition, we examined how the factors that emerged in our study were associated with several internalizing symptoms and with affect, as rated repeatedly in daily life. Such information is critical in determining the utility and validity of groupings of ER strategy use because factors are of little more than statistical interest if they are not related to meaningful relevant outcomes.

We evaluated the multilevel structure of ER strategies in two samples: a clinical sample using an ecological momentary assessment (EMA) design and an unselected student sample using a daily dairy design. As reviewed previously, there is little empirical work on the structure of ER at the between-person level (i.e., Aldao & Nolen-Hoeksema, 2010; Lee et al., 2015; Naragon-Gainey et al., 2017; Seligowski & Orcutt, 2015), and no studies have examined this structure using a repeated, context-specific approach or examined intraindividual structures over time. Thus, we elected to take an exploratory approach, given insufficient empirical evidence to specify confirmatory factors. The resulting exploratory structure was compared across samples and interpreted in light of the theoretical models and empirical studies reviewed previously.

Method

Participants

Student sample

A total of 315 participants were drawn from a large northeastern university’s undergraduate population.² Of the 315 participants who participated in the baseline assessment, 136 accepted the invitation to participate in the daily diary portion of the study. However, data from 27 participants were omitted from analyses because of extensive missing daily reports (i.e., missing > 70%) or poor-quality data (e.g., random or inconsistent responding, completing daily surveys outside of the required timeframe), leaving 109 participants for inclusion in analyses. This final sample had a mean age of 19.2 (SD = 1.8), was 56% female, and was racially diverse (50% White, 28% Asian or Pacific Islander, 10% Latino or Hispanic, 7% African American, 6% of other ethnicities). There were no differences in race/ethnicity between those whose data were included and those who were excluded from the daily diary analyses (p > .05). However, males were more likely to be excluded from the daily diary analyses (χ² = 12.147, df = 1, p < .001). For those participants included in analyses, an average of 83% of the daily surveys (1,259 total observations) were completed.

Clinical sample

A total of 163 treatment-seeking adults in the local community participated in a study on emotions. Participants were recruited using online and newspaper advertisements and flyers posted around the local community. They were eligible for the study if they were (a) currently receiving or seeking mental health treatment, (b) had a smartphone that could be used in the EMA study, and (c) did not report or exhibit current psychosis, dementia, or a cognitive impairment. Of the 163 participants from the baseline lab assessment, 141 (87%) participated in the EMA portion of the study. However, data from 12 participants were omitted from analyses because of extensive missing data (i.e., > 70% of EMA reports or less than three event-contingent reports; n = 11) or frequent invalid responding (i.e., completing most reports very quickly with no variability in responses; n = 1), leaving 129 participants for inclusion in analyses. There were no differences on any demographic variables (i.e., age, gender, race/ethnicity, or education level) between those who were included and those who were excluded from analyses (ps > .05). For those participants included in analyses, an average of 78% of the 10 event-contingent reports assessing ER strategy use and an average of 80% of the 30 EMA reports assessing symptoms and affect were submitted and appeared to be valid (i.e., not completed extremely quickly and submitted within 2 hr of the time the text was sent). Thus, there were 1,009 event-contingent reports, but note that only a subset (i.e., 933) of the EMA reports were used because each event-contingent report was linked to the next EMA report in order to test lagged within-person effects (see “Data Analysis”).

The final sample had a mean age of 30.4 years (SD = 11.9 years, range = 18–65), was moderately racially diverse (70% White, 15% African American, 10% Asian or Pacific Islander, 9% Latino or Hispanic, and 2% American Indian; multiple categories possible), and the majority were female (71%). The sample was fairly well educated: 40.5% had completed a 4-year degree or higher degree, 35.5% had some college education, 11% with a 2-year degree, and 10% with a high school diploma or some high school education. In terms of employment, 45% were employed full- or part-time, 44% were full- or part-time college or graduate students, 28% were unemployed, and 5% reported they were retired or did not need a job (multiple responses possible). Most of the sample stated that they were currently receiving therapy (67%) and/or taking medication (58%) for a psychological concern at the time of the study. According to the Anxiety Disorder Interview Schedule for DSM–5 (Brown & Barlow, 2014) administered at the lab baseline session, the most common diagnoses were generalized anxiety disorder (50%), social anxiety disorder (44%), persistent depressive disorder (30%), panic disorder (20%), and major depressive disorder (18%).

Procedure

Student sample

Participants came to the lab for the baseline session, in which they completed self-report measures and a computer task assessing executive function that is not relevant to the current study. They were compensated with course credit for their introductory psychology course. One day after attending the baseline session, participants received an e-mail inviting them to start the daily diary portion of the study from their homes using SurveyMonkey.com. Participants had 2 weeks to initiate participation before they were no longer eligible to participate, and they began the study on average 3 days (SD = 4.3) after completing the baseline assessment in the lab. Participants were then e-mailed a survey every day at 5 p.m. for 14 days and were instructed to complete the survey between 5 p.m. and 2 a.m. the day it was received. These surveys asked the respondents about their current affect and recent symptoms as well as their strongest emotional experience that day. For each survey completed within this timeframe, participants received $2, plus a $5 bonus if they completed all the daily surveys within each week. In addition, for each survey completed, participants were entered once into a lottery to win one of three $50 Amazon gift cards. Thus, participants could receive up to $38 and a $50 Amazon gift card.

Clinical sample

Participants in the clinical sample completed a 3- to 4-hr baseline session, which consisted of self-report measures, a computer task assessing executive function, and a semi-structured diagnostic interview (none of which are analyzed in the current study). Participants were compensated with $40 for completing the baseline session. At the end of the lab study, interested participants enrolled in the follow-up 10-day EMA study, and the research assistants explained the study procedure and example items to the participants. Beginning 1 day after the baseline study and continuing for 10 days in total, participants were sent text messages via the service SurveySignal (Hofmann & Patel, 2015). Each text message contained a link to a study questionnaire to be completed on the participant’s smartphone within 1 hr of receipt (though surveys completed within 2 hr were included in analyses). Three surveys assessing affect and symptoms were sent per day (30 in total). Two of these surveys were sent randomly between 9:30 a.m. and 5 p.m., with a minimum of 30 min in between the two surveys. The third survey was sent at 9 p.m. each evening; this evening report also included an assessment of positive and negative events not analyzed in the present study. In addition, participants were asked to self-initiate a daily event-contingent report prompted by the occurrence of a relatively strong positive or negative emotional experience, in which they provided information about their use of a number of different ER strategies during that emotional episode as well as other features of that experience that are not relevant to the current study. Participants were compensated $1 for every EMA questionnaire they completed, and if they missed no more than two questionnaires between Days 1 and 5 and/or between Days 6 and 10, they received a $5 bonus for each 5-day period. Thus, depending on how many surveys they completed, participants were compensated up to $50 for the EMA portion of the study.

Measures

ER strategies

For their strongest daily emotional experience, participants were instructed to rate their experience of positive and negative emotions at that time and then asked, “How did you respond to the strongest emotion you felt during this situation?” Participants were provided with single-item statements for each of 12 ER strategies (shown in Table 1), and they indicated whether they used each strategy by selecting yes or no. We selected strategies that are commonly studied in the clinical literature (e.g., in reviews or meta-analyses, theoretical models, or structural analyses; Aldao et al., 2010; Gross, 2015; Lee et al., 2015; Naragon-Gainey et al., 2017; Seligowski & Orcutt, 2015) and also used relatively frequently in daily life (e.g., Brans et al., 2013). Some of these items (i.e., distraction, expressive suppression, reappraisal, reflection, rumination, and seeking social support) have been used in other daily diary or EMA studies investigating ER strategy use (see Brans et al., 2013). For other strategies, we adapted items from retrospective self-report measures.

Table 1.

Means, Standard Deviations, and Intraclass Correlation Coefficients for Emotion-Regulation Strategies

		Clinical			Student
Strategy	Item	M	SD	ICC	M	SD	ICC
Acceptance	“I accepted the way I was feeling without judging myself”	.61	.49	.469	.56	.49	.332
Behavioral avoidance	“I avoided the situation that led to my feelings”	.21	.41	.303	.23	.42	.360
Distraction	“I engaged in activities to distract myself from my feelings”	.43	.49	.276	.34	.47	.327
Experiential avoidance	“I tried to get rid of negative thoughts, feelings, or sensations”	.49	.50	.396	.46	.49	.396
Expressive suppression	“I avoided expressing my feelings”	.36	.48	.298	.31	.46	.306
Procrastination	“I put off doing something that needed to get done”	.32	.47	.386	.34	.47	.354
Reappraisal	“I changed the way I thought about what caused my feelings”	.25	.44	.497	.23	.43	.377
Reflection	“I calmly reflected on my feelings and how to solve my problems”	.42	.49	.519	.34	.47	.323
Rumination	“I couldn’t stop thinking about my feelings (how bad I felt, why I felt that way)”	.40	.49	.350	.33	.47	.337
Savoring	“I tried to savor and hold onto my positive feelings”	.47	.49	.463	.35	.48	.238
Social support	“I talked about my feelings with others”	.27	.45	.334	.32	.47	.355
Substance use	“I used alcohol or other substances to change my feelings”	.05	.22	.329	.04	.19	.443

Positive and negative affect

For the daily diary reports in the student sample and each EMA report in the clinical sample, participants completed items from the Positive and Negative Affect Schedule (PANAS; Watson, Clark, & Tellegen, 1988) to assess current mood. Participants were asked, “To what extent do you feel the following emotions currently; that is, right now” and provided with select items from both the positive affect (PA) and negative affect (NA) subscales of the PANAS to rate on a scale of 1 (very slightly or not at all) to 5 (extremely). In the student sample, six highly loading PA items (i.e., excited, strong, active, interested, enthusiastic, attentive) and five highly loading NA items (i.e., upset, afraid, irritable, guilty, and distressed; Watson et al., 1988) were administered. In addition, sad was added to cover this component of negative mood. The mean Cronbach’s α for the NA scale across days was .88 (.83–.92), and the mean Cronbach’s α for the PA scale across days was .89 (.86–.92). In the clinical sample, four items were selected that provide broad coverage of aspects of PA (i.e., excited, strong, active, interested) and NA (i.e., upset, sad, afraid, irritable). The mean Cronbach’s α for the NA scale across days was .85 (.78–.91), and the mean Cronbach’s α for the PA scale across days was .83 (.74–.87).

Internalizing symptoms

For the daily diary reports in the student sample and each EMA report in the clinical sample, participants completed subscales of the Inventory of Depression and Anxiety Symptoms (IDAS; Watson et al., 2007). In the student sample, the instructions stated “Please select the answer that best describes your overall experience today (not just during the above emotional situation)” on a scale of 1 (not at all) to 5 (extremely), and participants were provided with five highly loading items from the IDAS Dysphoria (e.g., “I felt depressed”) and Social Anxiety (e.g., “I felt self-conscious knowing others were watching me”) scales. In the clinical sample, the instructions stated, “To what extent does each statement describe how you have felt within the past hour; that is, very recently” on a scale of 1 (not at all) to 5 (extremely) and were provided with four highly loading items from the IDAS Dysphoria, Social Anxiety, Panic (e.g., “I felt dizzy or lightheaded”), and Worry (e.g., “I felt nervous”) scales.

The IDAS has demonstrated good internal consistency and validity in both undergraduate and psychiatric samples (e.g., Watson et al., 2007, 2008). In the student sample, the mean Cronbach’s α for the Dysphoria subscale across days was .88, ranging from .83 to .93, and the mean α for the Social Anxiety subscale across days was .92, ranging from .88 to .95. In the clinical sample, the mean Cronbach’s αs and ranges for the IDAS symptoms subscales across days were as follows: Dysphoria mean α = .89 (.85–.92), Social Anxiety mean α = .89 (.84–.95), Worry mean α = .89 (.86–.92), and Panic mean α = .83 (.73–.91).

Data analysis

Because of the nested nature of the data (i.e., assessment occasions nested within each individual), data were analyzed using multilevel structural equation modeling in Mplus 8.0 (L. K. Muthén & Muthén, 2017). Mplus handles multilevel data by separating observed variables containing both between- and within-person variance into two uncorrelated latent variables. Thus, one latent variable represents the between-person variance in the observed variable, controlling for its within-person variance, and the other latent variable represents within-person variance, controlling for between-person variance. As part of the decomposition, the within-person latent variable is implicitly centered around each person’s mean. Because of the binary response format of the ER strategy items, mean- and variance-adjusted weighted least squares estimators (WLSMV) were used to appropriately model categorical indicators.

Exploratory factor analyses

Multilevel exploratory factor analyses were conducted in both samples to determine the number and nature of factors that best reproduce the observed within- and between-person covariance structures of daily ER strategies. An oblique geomin rotation was used, given that we expected the factors to be correlated. Several criteria were considered when determining the optimal number of factors to extract in each sample, including the scree plot, propriety of model solutions, model fit, and interpretability of factors.³ Model fit was assessed with the chi-square statistic, comparative fit index (CFI), Tucker-Lewis index (TLI), root-mean-square error of approximation (RMSEA), and standardized root-mean-square residual (SRMR) estimated separately for each level of the model. We followed the recommendations of Hu and Bentler (1999) for interpreting the approximate fit indices. According to these standards, CFI and TLI values greater than .90 indicate acceptable model fit, whereas values greater than .95 indicate excellent model fit. For RMSEA, values below 0.10 suggest acceptable model fit, whereas values below 0.06 indicate good model fit. For SRMR, values close to 0.08 or lower indicate acceptable fit.

Factor associations with affect and symptoms

After determining the optimal between- and within-person structure for ER strategies, we examined the associations of the factors with symptoms and affect. Given the model complexity and relatively small sample size, we used Bayesian structural equation modeling (BSEM), which is a recent alternative to traditional maximum likelihood estimation approaches that allows for more flexible parameter estimation and specification.⁴ Bayesian estimation can accommodate complex models (e.g., many cross-loadings) that are not tenable with other estimators, aided by researcher-specified prior estimates (i.e., priors) of the variances of the parameter distributions. Consistent with the guidelines of B. Muthén and Asparouhov (2012), we used noninformative priors for variances of primary factor loadings (i.e., loadings greater than |.40| in the EFA) and an informative prior (standardized) variance of .01 for all other factor loadings, which allows for cross-loadings with a mean of 0 and a 95% credibility interval ranging from −0.20 to +0.20. Model fit is evaluated through posterior predictive checking in BSEM, which focuses on the posterior predictive p value (PPP). The PPP refers to the proportion of generated data in which the posterior predictive likelihood ratio test is larger than the model test statistic, with values of .5 indicating good model fit, which suggests that the observed and generated data are equally probable. Although there is limited work suggesting the lower “cutoff” for good fit using PPP, values greater than .01 or .05 are considered acceptable. In addition, the 95% confidence interval of the PPP should contain zero (B. Muthén & Asparouhov, 2012; L. K. Muthén, Muthén, & Asparouhov, 2017).

In both samples, we included day of study participation as a covariate to account for linear trends over time. The assessment of ER strategy use, current affect, and daily symptoms were within a single report in the student sample, but affect and symptoms were assessed separately from strategy use in the clinical sample. Thus, we linked each daily event-contingent report of ER with the next EMA report assessing symptoms and affect (such that only one EMA report per participant per day was analyzed), and we included the time interval between the event-contingent and EMA reports (i.e., timelag) as a covariate in the clinical sample. To help account for systematically missing daily or EMA reports, forgetfulness (i.e., “How often do you forget to complete everyday tasks?”) was specified as an auxiliary variable, though it was not included as a parameter in the model itself.

Results

Means, standard deviations, and ICCs of the daily ER strategies are presented in Table 1. Because of its low base rate (endorsed in about 5% of submitted reports) in both samples, substance use was dropped from all analyses. In the clinical sample, ICCs ranged from .28 (distraction) to .52 (reflection), with a mean ICC of .39. In the student sample, ICCs ranged from .24 (savoring) to .39 (experiential avoidance), with a mean ICC of .34. The ICCs in both samples indicate that there was greater variability in ER strategy use across assessment occasions than across individuals, supportive of the idea that individuals select different strategies at different times.

Exploratory factor analyses

Table 2 presents the model fit indices for various combinations of one to four within- and between-factor solutions extracted from multilevel exploratory factor analyses in both the clinical and student samples. The table also shows model fit for each factor solution at each level, combined with unrestricted covariances at the other level (meaning that no factor structure is imposed on that level). We started by evaluating these models because they allow us to examine the fit of each level independent of the other.

Table 2.

Model Fit for Exploratory Factor Analyses of Emotion-Regulation-Strategy Use

Model (Factors)		Student				Clinical
Within-person	Between-person	χ² (df)	RMSEA	CFI/TLI	SRMR (w/b)	χ² (df)	RMSEA	CFI/TLI	SRMR (w/b)
Unrestricted	1	143.28** (44)	0.042	.922/.806	0/.171	204.19** (44)	0.060	.846/.616	0/.294
Unrestricted	2	61.76* (34)	0.025	.978/.930	0/.063	61.34** (34)	0.028	.974/.915	0/.065
Unrestricted	3	38.89* (25)	0.021	.989/.952	0/.042	28.49 (25)	0.012	.997/.985	0/.040
1	Unrestricted	618.37** (44)	0.102	.551/–.124	.126/0	508.98** (44)	0.102	.554/–.116	.156/0
2	Unrestricted	243.03** (34)	0.070	.836/.471	.073/0	128.62** (34)	0.053	.909/.706	.063/0
3	Unrestricted	129.74** (25)	0.058	.918/.639	.051/0	31.587 (25)	0.016	.994/.972	.031/0
4	Unrestricted	53.65** (17)	0.041	.971/.814	.029/0	—	—	—	—
1	1	745.04** (88)	0.077	.486/.357	.126/.171	643.03** (88)	0.079	.467/.334	.156/.294
2	1	357.85** (78)	0.053	.781/.691	.073/.171	323.02** (78)	0.056	.765/.668	.063/.294
3	1	254.12** (69)	0.046	.855/.769	.051/.171	256.16** (69)	0.052	.820/.714	.031/.294
4	1	190.75** (61)	0.041	.898/.817	.029/.171	—	—	—	—
1	2	792.12** (78)	0.085	.441/.212	.126/.063	669.45** (54)	0.087	.432/.199	.156/.065
2	2	339.67** (68)	0.056	.787/.656	.073/.063	199.11** (64)	0.044	.874/.796	.063/.065
3	2	208.004** (59)	0.045	.883/.783	.051/.063	81.720 * (59)	0.02	.978/.959	.031/.065
4	2	117.931** (51)	0.032	.948/.887	.029/.063	—	—	—	—
1	3	762.11** (69)	0.089	.458/.135	.126/.042	631.54** (69)	0.09	.460/.139	.156/.040
2	3	321.52** (59)	0.059	.795/.617	.073/.042	176.54** (59)	0.044	.887/.790	.063/.040
3	3	190.85** (50)	0.047	.890/.758	.051/.042	59.93 (50)	0.014	.990/.979	.031/.040
4	3	99.32** (42)	0.033	.955/.883	.029/.042	—	—	—	—

Note: Based on fit indices, as well as propriety and interpretability of model solutions, the three within-person and two between-person factor structure in the clinical sample and the four within-person and two between-person structure in the student sample (bold in the table) were determined to be the best solutions (see text for further detail). RMSEA = root-mean-square error of approximation; CFI = comparative fit index; TLI = Tucker-Lewis index; SRMR = standardized root-mean-square residual; (w/b) = within-person/between-person.

p < .05. **p < .01.

Clinical sample

According to the fit of models where structure was imposed at one level and unrestricted at the other level, only the three-factor solution at the within-person level evidenced good fit across all fit indices, whereas both the two- and three-factor solutions showed acceptable fit at the between-person level. Next, we evaluated the fit of the combined factor solutions for the three within- and two between-person factors as compared to the three within- and three-between factors models. On examination of the three within- and three between-person factor solution, we discovered a significant negative residual variance for distraction at the between-person level, suggesting overextraction and an improper solution that was not viable for further consideration. In contrast, the three within- and two between-factor model converged on a proper solution. Although the model chi-square was significant (χ² = 81.720, df = 59, p = .0268), the approximate fit indices all suggested that this model fit the data well. Furthermore, the two-factor between-person structure was supported by the scree plot (see Figure S1 in the Supplemental Material available online), and we felt that this solution was interpretable (see the following for detail), with both factors containing several primary indicators with loadings > |.40| and few indicators cross-loading on both factors.

Table 3 presents the standardized factor loadings for the best-fitting between-person factor structure in each sample, and Table 4 presents the standardized factor loadings for the best-fitting within-person factor structure in each sample. The between-person structure reflects covariance patterns across different people in their daily ER strategy use, akin conceptually to a cross-sectional factor analysis but built up from repeated assessments. In contrast, the within-person structure reflects covariance patterns in an individual’s ER strategy use across different occasions; that is, it highlights strategies people tend to use concurrently on a given occasion. In the clinical sample, the first between-person factor, which we labeled Avoidance, was composed of putatively maladaptive ER strategies that involve avoidance of emotional stimuli. As such, Avoidance was strongly defined by behavioral avoidance, expressive suppression, distraction, rumination, and procrastination (standardized loadings = .68–.88), with cross-loadings from experiential avoidance (.56) and (low) acceptance (–.40). The second factor, which we labeled Engagement, was primarily composed of putatively adaptive ER strategies that involve attending and directly responding to the emotion or emotion-eliciting stimulus. This factor was defined by savoring, reappraisal, reflection, social support, acceptance (loadings = .69–.89), and experiential avoidance (.60). There were also substantial cross-loadings from expressive suppression (–.43) and distraction (.34). Of note, experiential avoidance and distraction loaded positively and moderately on both factors. The Avoidance and Engagement factors were independent of one another (r = –.04).

Table 3.

Between-Person Standardized Factor Loadings and Standard Errors for Best-Fitting Factor Analyses of Emotion-Regulation-Strategy Use

	Clinical sample		Student sample
Indicator	1 Avoidance	2 Engagement	1 Engagement	2 Avoidance
Reflection	−.12 (.09)	.80* (.05)	.92* (.05)	.00 (.01)
Social support	−.06 (.11)	.72* (.08)	.61* (.11)	−.04 (.13)
Savoring	.01 (.02)	.89* (.04)	.68* (.10)	.08 (.11)
Reappraisal	.25* (.10)	.86* (.06)	.67 (.09)*	.34* (.09)
Acceptance	–.40* (.08)	.69* (.08)	.80* (.10)	−.35* (.11)
Experiential avoidance	.56* (.10)	.60* (.08)	.53* (.10)	.48* (.09)
Expressive suppression	.72* (.08)	–.43* (.11)	−.03 (.10)	.82* (.08)
Distraction	.68* (.06)	.34* (.12)	.31* (.09)	.68* (.07)
Procrastination	.68* (.08)	−.16 (.10)	.01 (.03)	.80* (.07)
Behavioral avoidance	.88* (.07)	.00 (.01)	.02 (.08)	.97* (.05)
Rumination	.68* (.07)	−.21 (.12)	−.11 (.10)	.93* (.06)

Note: Boldface type indicates factor loadings greater than .40. Values in parentheses are standard errors for factor loadings.

p < .05.

Table 4.

Standardized Factor Loadings and Standard Errors for Best-Fitting Factor Analyses of Emotion Regulation Strategy Use

	Clinical sample			Student sample
Indicator	1 Attentional shift	2 Avoidance	3 Emotional expression	1 Attentional shift	2 Acceptance	3 Avoidance	4 Emotional expression
Reappraisal	.82* (.06)	.01 (.01)	−.06 (.07)	.58* (.13)	−.01 (.05)	.15 (.23)	−.08 (.07)
Reflection	.62* (.06)	−.29* (.06)	−.04 (.08)	.76* (.08)	.02 (.03)	−.09 (.18)	.09 (.06)
Experiential avoidance	.59* (.05)	.19* (.05)	−.01 (.04)	.63* (.13)	−.06 (.06)	.23 (.24)	.02 (.04)
Acceptance	.32* (.06)	–.66* (.05)	.04 (.07)	.56* (.11)	.96* (.11)	.01 (.01)	−.01 (.01)
Savoring	.19* (.06)	–.74* (.04)	.01 (.02)	−.05 (.04)	.56* (.11)	−.15 (.19)	.22* (.06)
Distraction	.54 (.07)*	.39 (.06)*	.02 (.04)	.16* (.08)	.02 (.04)	.65* (.09)	−.04 (.04)
Rumination	−.01 (.01)	.88* (.04)	.11 (.08)	.01 (.03)	−.25* (.08)	.53* (.09)	.21* (.05)
Procrastination	.01 (.06)	.48* (.06)	.08 (.08)	−.05 (.05)	.26* (.07)	.68* (.09)	.04 (.03)
Behavioral avoidance	−.02 (.07)	.46* (.06)	.01 (.07)	.01 (.05)	−.01 (.05)	.72* (.07)	−.02 (.05)
Expressive suppression	.01 (.01)	.41* (.09)	–.53* (.20)	.03 (.03)	−.04 (.05)	.58* (.09)	–.45* (.08)
Social support	.36* (.16)	.00 (.01)	.91* (.33)	.09 (.09)	−.02 (.03)	.03 (.02)	.84* (.12)

Note: Boldface type indicates factor loadings greater than .40. Values in parentheses are standard errors for factor loadings.

p < .05.

At the within-person level, we labeled the first factor Attentional Shift because it was defined by strategies that involve attempts to ignore or disengage from the emotional experience (i.e., experiential avoidance at .59, distraction at .54) and attend to it in a way that alters one’s perspective on it (i.e., reappraisal at .82, reflection at .62). There were also some secondary loadings from acceptance (.32) and social support (.36). We labeled the second factor Avoidance, which had similar content as the Avoidance between-person factor, although it was more strongly marked by negative perseverative thought and more weakly by behavioral/expressive strategies and distraction. Specifically, the strongest loadings were for rumination, (low) savoring, (low) acceptance, procrastination, behavioral avoidance, and expressive suppression (|.41| to |.88|). We labeled the third factor Emotional Expression because it was composed of indicators reflecting the outward expression or communication of emotional experience (i.e., social support at .91 and [low] expressive suppression at –.53). Although this factor had only two primary indicators, both of which had cross-loadings on one of the other factors, we elected to retain it because of its interpretability, strong fit, and the fact that the within-person structure with two factors did not provide an acceptable fit to the data (see Table 2). Attentional Shift was uncorrelated with both Avoidance and Emotional Expression (rs = .07 and –.23, respectively; ps > .05), whereas Avoidance and Emotional Expression were weakly negatively correlated (r = –.26; p < .05).

Student sample

For the between-person solutions with unrestricted within-person covariances, both the two- and three-factor structures demonstrated acceptable fit across approximate fit indices, and chi-square values were approaching nonsignificance (Table 2). Examining the within-person solutions with unrestricted between-person covariances, the four-factor solution evidenced the best fit across approximate fit indices, although the fit of the three-factor solution was also acceptable. Next, we evaluated the fit of the factor solutions with simultaneously estimated within- and between-person structures. The only solutions with acceptable model fit for all indices were the four within- and two between-person and the four within- and three between-person factor structures. However, the four within- and three between-person solution had a negative residual variance for one of the indicators on the between-person level, suggesting overextraction at the between-person level. In addition, the two between-person factor structure was supported by the scree plot (see Figure S1 in the Supplemental Material) and appeared interpretable as it closely resembled the structure we obtained in the clinical sample. Therefore, we retained the four within- and two between-person factor solution for the student sample.

Tables 3 and 4 present the factor loadings for the multilevel factor structure of daily ER strategies in the student sample. At the between-person level, the first factor closely resembled the Engagement factor in the clinical sample, with primary loadings from reflection, acceptance, savoring, reappraisal, social support, and experiential avoidance (loadings = .53–.92) and a cross-loading from distraction (.31). The second factor closely resembled the Avoidance factor in the clinical sample. This factor had primary loadings from behavioral avoidance, rumination, expressive suppression, procrastination, distraction, and experiential avoidance (loadings = .48–.97) and two secondary cross-loadings from (low) acceptance (–.35) and reappraisal (.34). Again, experiential avoidance and distraction had moderate to large positive loadings on both factors. However, whereas these two factors were uncorrelated in the clinical sample, Engagement and Avoidance were moderately and positively correlated in the student sample (r = .39).

At the within-person level, the first factor was similar to the Attentional Shift factor in the clinical sample because it was defined by reflection (.76) and included primary loadings from experiential avoidance (.63) and reappraisal (.58) and a cross-loading from acceptance (.56), although it differed from the Attentional Shift factor in the clinical sample in that distraction (.16) did not load strongly on this factor. We labeled the second factor Acceptance because it was primarily defined by acceptance (.96) and savoring (.56). This factor appears to reflect holding onto and valuing one’s emotional experience. The third factor is similar to the Avoidance factor in the clinical sample, defined by behavioral avoidance, procrastination, distraction, expressive suppression, and rumination (loadings = .53–.72). However, in the clinical sample, low acceptance and low savoring loaded on this factor, whereas they split off to form a separate factor in the student sample. The fourth factor closely resembled the Emotional Expression factor in the clinical sample, indicating that this factor may be robust despite few loadings. It was again primarily defined by social support (.84) and a cross-loading from (low) expressive suppression (–.45). Attentional Shift was uncorrelated with Acceptance (r = –.25) and moderately positively correlated with Avoidance (r = .45), Acceptance was moderately negatively correlated with Avoidance (r = –.49), and Emotional Expression was uncorrelated with all factors in the student sample (rs = –.09 to .10).

Tests of model robustness

Factor comparisons across samples

To examine structural robustness, we used Tucker’s congruence coefficients as a quantitative index of factor similarity between samples. Tucker’s congruence coefficient (ϕ) is a standardized measure of factor loading similarity that represents the cosine of the angle between two vectors of loadings (Lorenzo-Seva & ten Berge, 2006; Tucker, 1951). Simulation work suggests that ϕ = 0.85 to 0.94 indicates “fair” similarity between factor loadings and ϕ = 0.95 to 1.00 indicates “good” similarity (Lorenzo-Seva & ten Berge, 2006). We found evidence for good similarity across samples in the between-person factor loadings (Engagement ϕ = 0.96, Avoidance ϕ = 0.99). However, given that the two samples had different within-person factor solutions (three within-person factors in the clinical sample and four within-person factors in the student sample), we could not compare the Acceptance factor across samples, and we expected that the within-person Avoidance factor would not be similar because two indicators loading on this factor in the clinical sample (acceptance and savoring) split from this factor in the student sample to form the Acceptance factor. Consistent with this, within-person Avoidance was not very similar across samples when all indicators were included (ϕ = 0.75). However, when acceptance and savoring were removed from this calculation, the within-person Avoidance factor demonstrated fair similarity across samples (ϕ = 0.90), suggesting that these two indicators primarily drove the sample differences in the Avoidance factor. Congruence coefficients indicated good similarity for within-person Emotional Expression (ϕ = 0.96) and fair similarity for within-person Attentional Shift (ϕ = 0.89).

Sensitivity analyses of included indicators

To assess the extent to which structural results may be influenced by the specific indicators included, we conducted EFAs in each sample in which we systematically excluded indicators one at a time (except for factors with only two primary indicators, which were removed together; see the Supplemental Material for details). We then compared these solutions to the structures with all 11 indicators described previously (i.e., the two between- and three within-person factors in the clinical sample and two between- and four within-person factors in the student sample) and indexed the factor similarities with congruence coefficients. The fit of each of these models was acceptable (see Table A Supplemental Material), with the exception that some within-person models did not converge (one model in the clinical sample, four in the student sample) and therefore could not be examined. Table B in the Supplemental Material shows the resulting congruence coefficients for each factor and model, which consistently indicated good similarity at both the between- and within-person levels. The average congruence coefficients were as follows: between-person Engagement ϕ = 0.99 in both samples, between-person Avoidance ϕ = 0.98 in the clinical sample and ϕ = 0.99 in the student sample, within-person Avoidance ϕ = 0.96 for the clinical sample and ϕ = 0.98 in the student sample, Attentional Shift ϕ = 0.97 in the clinical sample and ϕ = 0.99 in the student sample, Emotional Expression ϕ = 0.97 in the clinical sample and ϕ = 0.99 in the student sample, and Acceptance ϕ = 0.98. Rumination was the only strategy that seemed to exert a nonnegligible influence on the solution in the clinical sample because the Avoidance and Attentional Shift factors were not highly similar to the original model (ϕ = 0.72 and 0.82), though the Emotional Expression remained the same (ϕ = 0.97). Removing rumination did not alter the factors in the student sample (ϕ = 0.98–.99).

Prediction of symptoms and affect from ER strategy factors

BSEM model specification and fit

We first ran fully latent structural regression models, specifying the affect and symptoms factors as latent variables (using items as indicators) with informative priors for their cross-loadings. However, PPPs obtained in these models were borderline acceptable in the clinical sample (i.e., PPPs = .02–.03) to poor in the student sample (i.e., PPPs = 0), potentially indicative of overly parameterized or under-identified models (L. K. Muthén et al., 2017). Thus, to reduce the number of parameters, we elected to use observed variables (i.e., summed scale scores) for symptoms and affect rather than latent variables. Table C in the Supplemental Material shows the zero-order correlations and descriptive statistics (i.e., means, standard deviations, ICCs) for the affect and symptom variables. Of note, the between-person correlations between negative affect and several symptoms (i.e., dysphoria, worry) were very large in magnitude (rs = .72–.89), suggesting substantial redundancy in these analyses. However, given that these variables were more moderately correlated at the within-person level (rs = .55–.66), we elected to use both outcomes in validating the structural ER models.

The overall model fit for the final BSEM models (i.e., within- and between-person ER strategy factors predicting observed within- and between-person affect and symptoms) was acceptable in both samples (clinical sample PPP = .320, 95% CI = [–.48.41, 84.23] for affect and PPP = .232, 95% CI = [–40.76, 147.33] for symptoms; student sample PPP = .037, 95% CI = [–5.37, 130.67] for affect and PPP = .037, 95% CI = [–.4.32, 145.88] for symptoms). All primary indicator factor loadings were significant (i.e., 95% credibility intervals did not contain 0) and consistent with the results of the EFAs in both samples (see Supplemental Table D for the BSEM ER factor loadings in each sample), with the exception that the 95% credibility intervals of the markers for the within-person Emotional Expression factors (i.e., social support and expressive suppression) contained zero in the clinical sample. This was likely due to parameter-level under-identification because Emotional Expression was composed of only two indicators, both of which had significant cross-loadings on other within-person factors. Thus, associations with this factor should be interpreted with caution.

Affect

Table 5 presents the median of the standardized posterior distribution estimates predicting between- and within-person daily affect from the between- and within-person factor structure of daily ER strategy use in both samples. Each estimate was computed holding constant the other ER factors in the model and the model covariates (i.e., day of report completion in both samples and the time lag between reports in the clinical sample). At the between-person level, Avoidance was significantly negatively related to between-person PA in both the clinical (median β = −0.35) and student (β = −0.46) samples and was positively related to between-person NA in both the clinical (β = 0.56) and student (β = 0.84) samples. Similarly, Engagement was significantly positively related to between-person PA in both the clinical (β = 0.53) and student (β = 0.48) samples and negatively related to between-person NA in both the clinical (β = −0.24) and student (β = −0.29) samples.

Table 5.

Regression Paths From Bayesian Structural Equation Models of Within- and Between-Person Emotion-Regulation Factors as Predictors of Affect and Symptoms

Predictor	Positive affect		Negative affect		Dysphoria		Social anxiety		Worry		Panic
Predictor	Mdn β	95% CI	Mdn β	95% CI	Mdn β	95% CI	Mdn β	95% CI	Mdn β	95% CI	Mdn β	95% CI
Between-person
Clinical sample
Engagement	.53###	.34, .69	−.24#	−.43, –.03	−.26##	−.45, –.06	−.07	−.27, .14	−.07	−.27, .14	−.08	−.29, .13
Avoidance	−.35##	−.53, –.14	.56###	.38, .72	.59###	.40, .74	.46###	.26, .62	.64###	.46, .78	.39##	.17, .58
Student sample
Engagement	.48##	.23, .71	−.29##	−.54, –.07	−.35##	−.58, –.15	−.31##	−.55, –.09
Avoidance	−.46###	−.70, –.21	.84###	.68, 1.01	.90###	.75, 1.07	.76###	.58, .94
Within-person
Clinical sample
Day	−.08##	−.15, –.02	.02	−.04, .09	−.04	−.09, .03	−.08#	−.14, –.01	−.02	−.08, .05	.07#	.01, .14
Timelag	.04	−.03, .11	−.02	−.09, .05	−.06	−.13, .01	−.03	−.09, .04	−.06	−.13, .01	−.04	−.11, .02
Attentional shift	0	−.09, .09	.02	−.08, .12	.04	−.06, .15	.03	−.07, .13	.03	−.07, .13	0	−.09, .11
Avoidance	−.18###	−.27, –.09	.35###	.26, .43	.39###	.30, .48	.11#	.01, .22	.30###	.22, .39	.20###	.11, .29
Emotional expression	.01	−.10, .12	.05	−.07, .16	.05	−.07, .21	.07	−.10, .25	.06	−.07, .21	.03	−.12, .17
Student sample
Day	−.16###	−.21, –.11	−.03	−.09, .02	−.05#	−.09, –.01	−.17###	−.22, –.12
Attentional shift	.15#	.01, .30	−.28###	−.53, –.12	−.40###	−.75, –.21	−.18##	−.38, –.03
Acceptance	.29###	.15, .45	−.14	−.31, .16	−.09	−.26, .31	−.07	−.20, .14
Avoidance	−.22#	−.42, –.02	.59###	.39, .99	.82###	.61, 1.34	.35###	.16, .65
Emotional expression	.09	−.04, .20	.07	−.07, .25	.05	−.10, .20	.05	−.07, .17

Note: Mdn β = median standardized posterior parameter estimate; 95% CI = 95% credibility interval, or the lower and upper limits of the estimates posterior distribution; timelag = time elapsed between reports. The pound signs represent the proportion of the posterior distribution that overlaps with zero: # < .05; ## < .01; ### < .001.

At the within-person level, Attentional Shift was unrelated to within-person PA and NA in the clinical sample. However, Attentional Shift was significantly positively related to PA (β = 0.15) and negatively related to NA (β = −0.28) in the student sample. Within-person Avoidance was significantly negatively related to within-person PA in both samples (clinical β = −0.18; student β = −0.22) and positively related to within-person NA in both samples (clinical β = 0.35; student β = 0.59). Emotional Expression was unrelated to both PA and NA, whereas Acceptance (which was a distinct factor in the student sample only) was significantly positively related to within-person PA (β = 0.29) and unrelated to within-person NA.

Symptoms

Table 5 presents the results of the BSEM models examining the prediction of between- and within-person internalizing symptoms (i.e., dysphoria and social anxiety in both samples; worry and panic in the clinical sample only) from the within- and between-person factor structure of daily ER strategy use. At the between-person level, Engagement was negatively related to dysphoria in both samples (clinical β = −0.26; student β = −0.35) and was negatively related to social anxiety symptoms in the student sample (β = −0.31) but unrelated to all other symptoms in the clinical sample. The between-person Avoidance factor was positively related to dysphoria (clinical β = 0.59; student β = 0.90) and social anxiety (clinical β = 0.46; student β = 0.76) in both samples. In the clinical sample, Avoidance was also positively related to worry (β = 0.64) and panic symptoms (β = 0.39).

At the within-person level, Attentional Shift was unrelated to all symptoms in the clinical sample but significantly negatively related to both dysphoria (β = −0.14) and social anxiety symptoms (β = −0.14) in the student sample. The within-person avoidance factor was significantly positively related to dysphoria (clinical β = 0.39; student β = 0.82) and social anxiety (clinical β = 0.11; student β = 0.35) in both samples. In the clinical sample, the within-person Avoidance factor was also significantly positively related to worry (β = 0.30) and panic symptoms (β = 0.20). Finally, both Emotional Expression and Acceptance (which is only relevant for the student sample) were unrelated to all symptoms in both samples.

Discussion

The purpose of this study was to explore the multilevel structure of daily ER strategy use to provide a framework for understanding the naturalistic implementation and associated outcomes. We examined this structure in two samples—one clinical and one student—using two types of intensive longitudinal designs (i.e., EMA and daily diary) and predicted within- and between-person symptoms and affect from the ER strategy factors. At the between-person level, a two-factor structure representing Engagement (e.g., acceptance, reappraisal, reflection, savoring, seeking social support) and Avoidance (e.g., low acceptance, behavioral avoidance, distraction, procrastination, rumination, suppression) emerged in both the student and clinical samples. A different structure emerged at the within-person level, with three factors in the clinical sample (i.e., Attentional Shift, Avoidance, Emotional Expression) and the addition of a fourth factor (i.e., Acceptance) in the student sample.

Structure of individual differences in strategy use

It is striking that the between-person structure of daily ER use was very similar across samples—despite different study designs and sample differences in levels of psychopathology—suggesting that this structure may be robust (though further replication is required given the novelty of this examination). Moreover, it is consistent with a large literature identifying individual differences in broad biobehavioral approach and avoidance systems (e.g., Watson, Wiese, Vaidya, & Tellegen, 1999) and generally aligns with the adaptive-maladaptive framework (e.g., Aldao et al., 2010), with the Engagement factor composed of putatively adaptive strategies and the Avoidance factor composed of putatively maladaptive strategies.

Although Naragon-Gainey and colleagues (2017) reported a three-factor structure of ER strategy use on the basis of meta-analytic correlations of dispositional assessments (which should resemble the between-person structure of the present investigation), there are similarities between the structures found in the two investigations. Most notably, strategies that marked the Avoidance factor in the present investigation were strong indicators of two highly correlated factors (r = .67) in the Naragon-Gainey et al. meta-analysis: Specifically, Disengagement in that study was marked by behavioral avoidance, distraction, expressive suppression, and experiential avoidance, and Aversive Cognitive Perseveration was marked by experiential avoidance, rumination, and low acceptance. In addition, the Engagement factor in the present investigation and the Adaptive Engagement factor reported by Naragon-Gainey et al. contained a strong loading from reappraisal, whereas the remaining indicators on each factor were putatively adaptive strategies that were not included in both studies and therefore cannot be compared across studies (i.e., 3 of 12 strategies were specific to the current study, whereas 1 of 10 was specific to the meta-analysis). Given that EFA results are strongly influenced by the content subjected to the analysis, we suspect that the minor differences between these two investigations were largely due to the specific strategies examined, though we cannot rule out the possibility that the observed differences reflect the assessment methods (momentary use vs. trait retrospective reports).

Importantly, each factor in both samples demonstrated theoretically consistent associations with both symptoms and positive and negative affect. That is, at the between-person level, overall use of Engagement strategies was associated with less internalizing symptoms and negative affect as well as higher positive affect. In contrast, the overall use of Avoidance strategies was positively associated with symptoms and negative affect and negatively associated with positive affect. Levels of reported Engagement and Avoidance were unrelated in the clinical sample such that tendencies to use one group of strategies was not informative about tendencies to use the other group. However, the positive correlation between these factors in the student sample likely represents individual differences in the tendency to regulate emotions in any manner because a subgroup of “low regulators” is likely to be larger in this less distressed sample (Dixon-Gordon et al., 2015; see also mean levels of strategy use in Table 1).

Structure of intraindividual momentary strategy use

Although the within-person factor structure was also generally similar across samples, there are some important differences that should be considered when interpreting the results. The primary difference was that Acceptance (composed solely of acceptance and savoring) emerged as its own factor in the student sample, whereas these strategies loaded negatively onto the Avoidance factor in the clinical sample. Given that savoring is by definition relevant to positive emotions only and individuals may be more likely to use acceptance in response to positive emotions than negative emotions, more frequent experiences of positive affect may have led to the emergence of an Acceptance factor in the student sample. An additional difference between the within-person structures is that distraction loaded on both Attentional Shift and within-person Avoidance in the clinical sample, whereas it is a primary marker for within-person Avoidance alone in the student sample. The sample differences in negative affect intensity may also explain why distraction shares a cross-loading with Attentional Shift in the clinical sample but not the student sample because distraction has been shown to be more effective and preferable to other cognitively demanding strategies (e.g., reappraisal) as negative affect intensity increases (e.g., Shafir, Schwartz, Blechert, & Sheppes, 2015; Sheppes et al., 2011; Sheppes & Meiran, 2008).

Similarly, deficits in attentional processing have been implicated in the relationship between ER strategies (e.g., reappraisal, rumination) and internalizing symptoms (e.g., Hsu et al., 2015; Pe et al., 2013). It is possible that the clinical sample may have employed distraction more indiscriminately as a result of decreased reliance on executive functions (e.g., McRae et al., 2010), leading to less effective implementation of the remaining Attentional Shift strategies. This is consistent with the finding that Attentional Shift was more weakly associated with affect and symptoms in the clinical sample than the student sample. Thus, Attentional Shift strategies may not be as feasible or effective for clinically distressed individuals, resulting in reliance on less effortful (and potentially less effective) strategies such as distraction.

Whereas within-person Avoidance demonstrated consistent positive associations with affect and symptoms, Emotional Expression was unrelated to symptoms and affect in both samples. This may be due to few primary loadings on this factor because social support and expressive suppression both had significant loadings on the Attentional Shift or Avoidance factors and relatively large confidence intervals. Although this factor did emerge in both samples, it will be important for future studies to examine whether it is robust by including other potential indicators (e.g., emotional expressivity itself, venting, sharing). Additionally, whereas the other within-person factors were composed of a mix of overt and covert strategies, Emotional Expression was defined solely by overt strategies involving displaying (or hiding) one’s emotions. It is possible that the adaptiveness of these strategies may be particularly dependent on the social context. For example, some situations may call for the active suppression of affective components (e.g., receiving negative feedback from an employer), whereas other situations may require expressing one’s feelings (e.g., addressing an issue with a spouse), and the social consequences of its use may determine how adaptive or effective it is. Consistent with this idea, expressive suppression is associated with numerous social outcomes (e.g., decreases in receiving social support, decreases in likability), whereas more covert strategies such as reappraisal are not (e.g., English & John, 2013; Srivastava, Tamir, McGonigal, John, & Gross, 2009).

Given that the within-person structure of the present investigation models the momentary covariance of strategy use, it is worth considering how it aligns with the different temporal stages of ER proposed by Gross (1998, 2015) in the process model. To some degree, Attentional Shift is similar to the cognitive change strategies proposed by Gross, and both Emotional Expression strategies correspond to response modulation strategies. However, the Avoidance factor spans numerous groupings in the process model, including situation selection, attentional deployment, and response modulation. Discrepancies between the theoretical temporal occurrence of strategies proposed by Gross and their actual occurrence also have been demonstrated in previous studies (e.g., Kalokerinos, Résibois, Verduyn, & Kuppens, 2017; Paul, Simon, Kniesche, Kathmann, & Endrass, 2013), but it is important to note that our study did not have the temporal resolution needed to optimally test the process model (i.e., the assessment of strategy use over the course of seconds or minutes and in real time).

Clinical implications

The present study is the first to empirically demonstrate that patterns in the ways strategies are selected on a given occasion may influence an individual’s subsequent symptoms and affect. Although this was assessed concurrently in the student sample, the associations were demonstrated prospectively in the clinical sample. Within-person prospective associations are consistent with (although do not prove) a causal role of ER strategy use on affective symptoms and experiences. Furthermore, if associations of ER with symptoms were solely a result of confounding individual differences that can affect retrospective reporting (e.g., personality traits like neuroticism, level of psychopathology; Clark, Vittengl, Kraft, & Jarrett, 2003), we would not see meaningful within-person variability based on strategy use, which was observed here. These findings also increase confidence in the potential utility of interventions focused on ER for internalizing symptoms and well-being (e.g., dialectical behavioral therapy, acceptance and commitment therapy). In particular, our results suggest that it may be helpful to assess and then track usage of ER strategies during therapy according to the within-person factor groupings. For those with more severe symptoms, it may be best to initially focus on reducing avoidance-related strategies; in later stages of therapy or in more mild cases, it may be particularly useful to work to incorporate strategies on the Attentional Shift factor. These results are consistent with other work indicating that putatively maladaptive strategies have more impact on symptoms than putatively adaptive strategies in clinical samples (Aldao, Jazaieri, Goldin, & Gross, 2014; Plate, Aldao, Quintero, & Mennin, 2016).

It is notable that in both the between- and within-person structures and across samples, distraction and experiential avoidance had cross-loadings on approach (i.e., Engagement, Attentional Shift) and avoidance (i.e., Avoidance) factors. These mixed associations are reflected in the larger literature: Experiential avoidance is intended to control or eliminate unwanted internal experiences (e.g., thoughts and emotions), but its habitual use tends to prolong and intensify such experiences in the long term (e.g., Hayes, Strosahl, & Wilson, 1999; Kashdan, Barrios, Forsyth, & Steger, 2006). Similarly, distraction has been proposed as a counterproductive ER strategy in some therapeutic protocols (e.g., Craske & Barlow, 2008), whereas others have argued that the distraction can be a temporarily useful strategy for managing intense distressing emotions (e.g., Linehan & Kehrer, 1993). In addition, other studies have reported that use of distraction covaries with both acceptance-based and avoidance-based strategies (Wolgast & Lundh, 2017). How do we reconcile these conflicting results? One possibility is that the presence of other strategies employed (or specific contextual features; see “Limitations and Future Directions”) on a given occasion determines the workability of using distraction or experiential avoidance. This interpretation is consistent with the growing emphasis in the ER literature on repertoires of strategies and interactions among them rather than consequences of individual strategy use (Aldao & Dixon-Gordon, 2014; Aldao et al., 2014; Aldao & Nolen-Hoeksema, 2012; Plate et al., 2016).

Limitations and future directions

There are a number of limitations in the present work that warrant consideration and should be addressed in future work. First, the methodology of the present study required individuals to select their strongest emotional experience and in the clinical sample, to initiate a report. Although compliance rates were good in both samples and other approaches (e.g., random fixed signals) are likely to miss important emotional experiences, this approach may result in systematic biases in selection or failure to report emotional episodes. In addition, different findings across samples may be due to differences in study design (i.e., daily diary in student sample and event-contingent/EMA in the clinical sample), so the content-based explanations of samples differences that we discussed are speculative. Given the overall structural robustness across samples, it does not appear that design had a strong and generalized different impact on the structural analyses. However, the retrospective recall inherent to the daily diary design in the student sample may have inflated the associations between ER factors and some outcomes (e.g., differences across samples in the associations of Attentional Shift with outcomes).

In addition, the binary response format assessing individual strategy use does not capture the extent to which individuals engaged in strategy use, which is primarily relevant to responses near the threshold between endorsing yes or no and in cases where there is substantial heterogeneity in the degree of use (e.g., within a yes response). We selected this because a more differentiated Likert scale response would be more burdensome for participants and could potentially reduce the number or quality of reports, but the binary response format may have led to over- or underreporting of some strategies. Similarly, while we included a relatively large number of commonly used and studied strategies and the results of our sensitivity analyses suggest that indicator inclusion did not strongly influence the factor structure, the strategies selected were not exhaustive. A different factor structure may have emerged if more or different ER strategies (e.g., mindfulness, active problem solving, emotional eating, self-harm) were included. Finally, the low base rate of substance use as an ER strategy is surprising given that coping is one common substance use motive (e.g., O’Hare & Shen, 2012) and ER deficits are associated with momentary substance use (e.g., Weiss, Bold, Sullivan, Armeli, & Tennen, 2017). It is possible that examining this factor structure in a sample with higher rates of substance use or with more substance use items would reveal a different structure.

This study is an initial step in illuminating how ER strategies are used in daily life and their consequences. However, there are numerous issues not examined here that will be important to consider in future studies. For example, we did not evaluate the influence of specific contextual variables (e.g., regulatory goals, intensity of affect, presence of other people) on strategy use. Thus, an important venture for future work on the momentary covariance of ER strategies is understanding contextual features that contribute to ER strategy choice and subsequent outcomes. Taking this a step further, a more nuanced way to understand the match between context and a given ER strategy is to examine the interactions among the person (e.g., personality, cognitive processes, emotional intelligence), context, and strategy (Doré, Silvers, & Ochsner, 2016). Further, individuals can have competing simultaneous ER goals (Tamir, 2016), which means they may select different strategies for different reasons within the same emotional episode. Understanding how these variables interact to predict differences in the selection and use of ER strategies will further elucidate the development of pathological ER as well as delineate points of intervention. Overall, we hope our work spurs additional research into the contextual and dynamic nature of how one responds to emotions and the outcomes of such responses.

Supplemental Material

McMahon_OpenPracticesDisclosure – Supplemental material for The Multilevel Structure of Daily Emotion-Regulation-Strategy Use: An Examination of Within- and Between-Person Associations in Naturalistic Settings

Supplemental material, McMahon_OpenPracticesDisclosure for The Multilevel Structure of Daily Emotion-Regulation-Strategy Use: An Examination of Within- and Between-Person Associations in Naturalistic Settings by Tierney P. McMahon and Kristin Naragon-Gainey in Clinical Psychological Science

Supplemental Material

McMahon_Supplemental_Material – Supplemental material for The Multilevel Structure of Daily Emotion-Regulation-Strategy Use: An Examination of Within- and Between-Person Associations in Naturalistic Settings

Supplemental material, McMahon_Supplemental_Material for The Multilevel Structure of Daily Emotion-Regulation-Strategy Use: An Examination of Within- and Between-Person Associations in Naturalistic Settings by Tierney P. McMahon and Kristin Naragon-Gainey in Clinical Psychological Science

Footnotes

Action Editor

John J. Curtin served as action editor for this article.

Author Contributions

Both authors contributed to the study conceptualization and design. Data collection and interpretation were performed by both authors, and T. P. McMahon performed the data analysis. T. P. McMahon drafted the manuscript, and K. Naragon-Gainey provided critical revisions. Both authors approved the final manuscript for submission.

ORCID iD

Tierney P. McMahon

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Supplemental Material

Additional supporting information can be found at

Open Practices

All data have been made publicly available via Open Science Framework and can be accessed at https://osf.io/pwy9r. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/2167702618807408. This article has received the badge for Open Data. More information about the Open Practices badges can be found at .

Notes

References

Aldao

(2013). The future of emotion regulation research: Capturing context. Perspectives on Psychological Science, 8, 155–172. doi:10.1177/1745691612459518

Aldao

Dixon-Gordon

K. L.

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

Aldao

Jazaieri

Goldin

P. R.

Gross

J. J.

(2014). Adaptive and maladaptive emotion regulation strategies: Interactive effects during CBT for social anxiety disorder. Journal of Anxiety Disorders, 28, 382–289. doi:10.1016/j.janxdis.2014.03.005

Aldao

Nolen-Hoeksema

(2010). Specificity of cognitive emotion regulation strategies: A transdiagnostic examination. Behaviour Research and Therapy, 48, 974–983. doi:10.1016/j.brat.2010.06.002

Aldao

Nolen-Hoeksema

(2012). The influence of context in the implementation of adaptive emotion regulation strategies. Behaviour Research and Therapy, 50, 493–501. doi:10.1016/j.brat.2012.04.004

Aldao

Nolen-Hoeksema

Schweizer

(2010). Emotion-regulation strategies across psychopathology: A meta-analytic review. Clinical Psychology Review, 30, 217–237. doi:10.1016/j.cpr.2009.11.004

Aldao

Sheppes

Gross

J. J.

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

Berking

Wupperman

(2012). Emotion regulation and mental health: Recent finding, current challenges, and future directions. Current Opinion in Psychiatry, 25, 128–134. doi:10.1097/YCO.0b013e3283503669

Bolger

Laurenceau

(2013). Intensive longitudinal methods: An introduction to diary and experience sampling research. New York, NY: Guilford Press.

10.

Bonanno

G. A.

Burton

C. L.

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

11.

Brans

Koval

Verduyn

Lim

Y. L.

Kuppens

(2013). The regulation of negative and positive affect in daily life. Emotion, 13, 926–939. doi:10.1037/a0032400

12.

Brockman

Ciarrochi

Parker

Kashdan

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

13.

Brown

T. A.

(2015). Confirmatory factor analysis for applied research (2nd ed.). New York, NY: Guilford Press.

14.

Brown

T. A.

Barlow

D. H.

(2014). Anxiety and related disorders interview schedule for DSM-5 (ADIS-5L): Client interview schedule. Oxford, UK: Oxford University Press.

15.

Clark

L. A.

Vittengl

Kraft

Jarrett

R. B.

(2003). Separate personality traits from states to predict depression. Journal of Personality Disorders, 17, 152–172. doi:10.1521/pedi.17.2.152.23990

16.

Compas

B. E.

Jaser

S. S.

Dunbar

J. P.

Watson

K. H.

Bettis

A. H.

Gruhn

M. A.

Williams

E. K.

(2014). Coping and emotion regulation from childhood to early adulthood: Points of convergence and divergence. Australian Journal of Psychology, 66, 71–81. doi:10.1111/ajpy.12043

17.

Craske

M. G.

Barlow

D. H.

(2008). Panic disorder and agoraphobia. In Barlow

D. H.

(Ed.), Clinical handbook of psychological disorders: A step-by-step treatment manual (4th ed., pp. 1–64). New York, NY: Guilford Press.

18.

Dixon-Gordon

K. L.

Aldao

De Los Reyes

(2015). Emotion regulation in context: Examining the spontaneous use of strategies across emotional intensity and type of emotion. Personality and Individual Differences, 86, 271–276. doi:10.1016/j.paid.2015.06.011

19.

Doré

B. P.

Silvers

J. A.

Ochsner

K. N.

(2016). Toward a personalized science of emotion regulation. Social and Personality Psychology Compass, 10(4), 171–187. doi:10.1111/spc3.12240

20.

English

John

O. P.

(2013). Understanding the social effects of emotion regulation: The mediating role of authenticity for individual differences in suppression. Emotion, 13(2), 314–329. doi:10.1037/a0029847

21.

English

Lee

I. A.

John

O. P.

Gross

J. J.

(2017). Emotion regulation strategy selection in daily life: The role of social context and goals. Motivation and Emotion, 41, 230–242. doi:10.1007/s11031-016-9597-z

22.

Gallagher

M. W.

(2017). Transdiagnostic mechanisms of change and cognitive-behavioral treatments for PTSD. Current Opinion in Psychology, 14, 90–95. doi:10.1016/j.copsyc.2016.12.002

23.

Gross

J. J.

(1998). The emerging field of emotion regulation: An integrative review. Review of General Psychology, 2, 271–299. doi:10.1037/1089-2680.2.3.271

24.

Gross

J. J.

(2013). Emotion regulation: Taking stock and moving forward. Emotion, 13, 359–365. doi:10.1037/a0032135

25.

Gross

J. J.

(2015). Emotion regulation: Current status and future prospects. Psychological Inquiry, 26, 1–26. doi:10.1080/1047840X.2014.940781

26.

Gross

J. J.

Jazaieri

(2014). Emotion, emotion regulation, and psychopathology: An affective science perspective. Clinical Psychological Science, 2, 387–401. doi:10.1177/2167702614536164

27.

Haines

S. J.

Gleeson

Kuppens

Hollenstein

Ciarrochi

Labuschagne

. . . Koval

. (2016). The wisdom to know the difference: Strategy-situation fit in emotion regulation in daily life is associated with well-being. Psychological Science, 27, 1651–1659. doi:10.1177/0956797616669086

28.

Hayes

S. C.

Strosahl

K. D.

Wilson

K. G.

(1999). Acceptance and commitment therapy: An experiential approach to behavior change. New York, NY: Guilford.

29.

Heiy

J. E.

Cheavens

J. S.

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

30.

Hofmann

Patel

P. V.

(2015). SurveySignal: A convenient solution for experience sampling research using participants' own smartphones. Social Science Computer Review, 33, 235–253. doi:10.1177/0894439314525117

31.

Hsu

K. J.

Beard

Rifkin

Dillon

D. G.

Pizzagalli

D. A.

Björgvinsson

(2015). Transdiagnostic mechanisms in depression and anxiety: The role of rumination and attentional control. Journal of Affective Disorders, 188, 22–27. doi:10.1016/j.jad.2015.08.008

32.

Bentler

P. M.

(1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling, 6, 1–55. doi:10.1080/10705519909540118

33.

Kalokerinos

E. K.

Résibois

Verduyn

Kuppens

(2017). The temporal deployment of emotion regulation strategies during negative emotional episodes. Emotion, 17, 450–458. doi:10.1037/emo0000248

34.

Kalokerinos

E. K.

Tamir

Kuppens

(2017). Instrumental motives in negative emotion regulation in daily life: Frequency, consistency, and predictors. Emotion, 17, 648–657. doi:10.1037/emo0000269

35.

Kashdan

T. B.

Barrios

Forsyth

J. P.

Steger

M. F.

(2006). Experiential avoidance as a generalized psychological vulnerability: Comparisons with coping and emotion regulation strategies. Behaviour Research and Therapy, 44, 1301–1320. doi:10.1016/j.brat.2005.10.003

36.

Koole

S. L.

(2009). The psychology of emotion regulation: An integrative review. Cognition and Emotion, 23, 4–41. doi:10.1080/02699930802619031

37.

Lee

D. J.

Witte

T. K.

Weathers

F. W.

Davis

M. T.

(2015). Emotion regulation strategy use and posttraumatic stress disorder: Associations between multiple strategies and specific symptom clusters. Journal of Psychopathology and Behavioral Assessment, 37, 533–544. doi:10.1007/s10862-014-9477-3

38.

Linehan

M. M.

Kehrer

C. A.

(1993). Borderline personality disorder. In Barlow

D. H.

(Ed.), Clinical handbook of psychological disorders: A step-by-step treatment manual (2nd ed., pp. 396–441). New York, NY: Guilford Press.

39.

Lorenzo-Seva

ten Berge

J. F.

(2006). Tucker’s congruence coefficient as a meaningful index of factor similarity. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 2, 57–64. doi:10.1027/1614-2241.2.2.57

40.

Maas

C. M.

Hox

J. J.

(2005). Sufficient sample sizes for multilevel modeling. Methodology: European Journal of Research Methods for the Behavioral and Social Sciences, 1, 86–92. doi:10.1027/1614-2241.1.3.86

41.

McMahon

T. P.

Naragon-Gainey

(2018). The moderating effect of maladaptive emotion regulation strategies on reappraisal: A daily diary study. Cognitive Therapy and Research, 42, 552–564.

42.

McRae

Hughes

Chopra

Gabrieli

J. E.

Gross

J. J.

Ochsner

K. N.

(2010). The neural bases of distraction and reappraisal. Journal of Cognitive Neuroscience, 22, 248–262. doi:10.1162/jocn.2009.21243

43.

Mehl

M. R.

Conner

T. S.

(2012). Handbook of research methods for studying daily life. New York, NY: Guilford Press.

44.

Moyal

Cohen

Henik

Anholt

G. E.

(2015). Emotion regulation as a main mechanism of change in psychotherapy. Behavioral and Brain Sciences, 38, e18. doi:10.1017/S0140525X14000259

45.

Muthén

Asparouhov

(2012). Bayesian structural equation modeling: A more flexible representation of substantive theory. Psychological Methods, 17, 313–335. doi:10.1037/a0026802

46.

Muthén

L. K.

Muthén

B. O.

(2017). Mplus user’s guide (8th ed.). Los Angeles, CA: Muthén & Muthén.

47.

Muthén

L. K.

Muthén

B. O.

Asparouhov

(2017). Regression and mediation analysis using Mplus. Los Angeles, CA: Muthén & Muthén.

48.

Naragon-Gainey

. (2019). Affective models of depression and anxiety: Extension to within-person processes in daily life. Journal of Affective Disorders, 243, 241–248.

49.

Naragon-Gainey

DeMarree

K. G.

(2017). Decentering attenuates the associations of negative affect and positive affect with psychopathology. Clinical Psychological Science, 5, 1027–1047.

50.

Naragon-Gainey

McMahon

T. P.

Chacko

T. P.

(2017). The structure of common emotion regulation strategies: A meta-analytic examination. Psychological Bulletin, 143, 384–427. doi:10.1037/bul0000093

51.

O'Hare

Shen

. (2012). Substance use motives and severe mental illness. Journal of Dual Diagnosis, 8, 171–179. doi:10.1080/15504263.2012.696184

52.

Parkinson

Totterdell

(1999). Classifying affect-regulation strategies. Cognition and Emotion, 13, 277–303. doi:10.1080/026999399379285

53.

Park

Naragon-Gainey

(2018). Daily experiences of emotional clarity and their association with internalizing symptoms in naturalistic settings. Emotion. Advance online publication. doi:10.1037/emo0000482

54.

Paul

Simon

Kniesche

Kathmann

Endrass

(2013). Timing effects of antecedent- and response-focused emotion regulation strategies. Biological Psychology, 94, 136–142. doi:10.1016/j.biopsycho.2013.05.019

55.

M. L.

Raes

Koval

Brans

Verduyn

Kuppens

(2013). Interference resolution moderates the impact of rumination and reappraisal on affective experiences in daily life. Cognition and Emotion, 27, 492–501. doi:10.1080/02699931.2012.719489

56.

Plate

A. J.

Aldao

Quintero

J. M.

Mennin

D. S.

(2016). Interactions between reappraisal and emotional nonacceptance in psychopathology: Examining disability and depression symptoms in generalized anxiety disorder. Cognitive Therapy and Research, 40, 733–746. doi:10.1007/s10608-016-9793-x

57.

Roesch

S. C.

Aldridge

A. A.

Stocking

S. N.

Villodas

Leung

Bartley

C. E.

Black

L. J.

(2010). Multilevel factor analysis and structural equation modeling of daily diary coping data: Modeling trait and state variation. Multivariate Behavioral Research, 45, 767–789. doi:10.1080/00273171.2010.519276

58.

Seligowski

A. V.

Orcutt

H. K.

(2015). Examining the structure of emotion regulation: A factor-analytic approach. Journal of Clinical Psychology, 71, 1004–1022. doi:10.1002/jclp.22197

59.

Shafir

Schwartz

Blechert

Sheppes

(2015). Emotional intensity influences pre-implementation and implementation of distraction and reappraisal. Social Cognitive and Affective Neuroscience, 10, 1329–1337. doi:10.1093/scan/nsv022

60.

Sheppes

Meiran

(2008). Divergent cognitive costs for online forms of reappraisal and distraction. Emotion, 8, 870–874. doi:10.1037/a0013711

61.

Sheppes

Scheibe

Suri

Gross

J. J.

(2011). Emotion-regulation choice. Psychological Science, 22, 1391–1396. doi:10.1177/0956797611418350

62.

Sheppes

Suri

Gross

J. J.

(2015). Emotion regulation and psychopathology. Annual Review of Clinical Psychology, 11, 379–405. doi:10.1146/annurev-clinpsy-032814-112739

63.

Srivastava

Tamir

McGonigal

K. M.

John

O. P.

Gross

J. J.

(2009). The social costs of emotional suppression: A prospective study of the transition to college. Journal of Personality and Social Psychology, 96, 883–897. doi:10.1037/a0014755

64.

Tamir

(2016). Why do people regulate their emotions? A taxonomy of motives in emotion regulation. Personality and Social Psychology Review, 20, 199–222. doi:10.1177/1088868315586325

65.

Tucker

L. R.

(1951). A method for synthesis of factor analysis studies (Personnel Research Section Report No. 984). Washington, DC: Department of the Army.

66.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070. doi:10.1037/0022-3514.54.6.1063

67.

Watson

O'Hara

M. W.

Chmielewski

McDade-Montez

E. A.

Koffel

Naragon

Stuart

(2008). Further validation of the IDAS: Evidence of convergent, discriminant, criterion, and incremental validity. Psychological Assessment, 20, 248–259. doi:10.1037/a0012570

68.

Watson

O'Hara

M. W.

Simms

L. J.

Kotov

Chmielewski

McDade-Montez

E. A.

. . . Stuart

. (2007). Development and validation of the Inventory of Depression and Anxiety Symptoms (IDAS). Psychological Assessment, 19, 253–268. doi:10.1037/1040-3590.19.3.253

69.

Watson

Wiese

Vaidya

Tellegen

(1999). The two general activation systems of affect: Structural findings, evolutionary considerations, and psychobiological evidence. Journal of Personality and Social Psychology, 76, 820–838. doi:10.1037/0022-3514.76.5.820

70.

Webb

T. L.

Miles

Sheeran

(2012). Dealing with feeling: A meta-analysis of the effectiveness of strategies derived from the process model of emotion regulation. Psychological Bulletin, 138, 775–808. doi:10.1037/a0027600

71.

Weiss

N. H.

Bold

K. W.

Sullivan

T. P.

Armeli

Tennen

(2017). Testing bidirectional associations among emotion regulation strategies and substance use: A daily diary study. Addiction, 112, 695–704. doi:10.1111/add.13698

72.

Wolgast

Lundh

(2017). Is distraction an adaptive or maladaptive strategy for emotion regulation? A person-oriented approach. Journal of Psychopathology and Behavioral Assessment, 39, 117–127. doi:10.1007/s10862-016-9570-x

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.60 MB

0.14 MB