Introduction
Distress tolerance refers to one’s perceived or actual ability to withstand or tolerate aversive emotions or physical states (Leyro et al., 2010; Zvolensky et al., 2010). Distress tolerance is often considered a development and maintenance factor for various kinds of internalizing and externalizing psychopathology (Leyro et al., 2010). There are generally two branches of research concerning distress tolerance. Behavioral distress tolerance concerns the actual behavioral capacity to tolerate aversive emotions or physical states, whereas perceived distress tolerance typically refers to the perceived or self-reported capacity to tolerate aversive emotional states or physical experiences (Zvolensky et al., 2010). Although both constructs refer to distress tolerance and are associated with psychopathology, perceived and behavioral distress tolerance are often uncorrelated or demonstrate only a small correlation with each other (Buckheit et al., 2020; Hsu et al., 2023; McHugh et al., 2011).
A more specific facet of the construct, perceived emotional distress tolerance, is one’s self-reported capacity to tolerate negative emotions (Simons & Gaher, 2005; Zvolensky et al., 2010). Although it has some construct overlap and correlations with other transdiagnostic processes like avoidant coping (e.g., Crabtree et al., 2019), experiential avoidance (e.g., Iverson et al., 2012), and emotion dysregulation (e.g., McHugh et al., 2013), it also relates to one’s self-efficacy and willingness to sit with distress rather than avoid (Veilleux, 2023). Self-reported emotional distress tolerance is expected to remain generally consistent over time, as a trait-like construct (Cummings et al., 2014; Kiselica et al., 2014; Leyro et al., 2010), although some research has also found that it can demonstrate change in the context of treatment (Heiland & Veilleux, 2024). As expected of a transdiagnostic factor, self-reported distress tolerance is associated with a range of psychological disorders and concerns, including posttraumatic stress disorder (PTSD), problematic substance use, major depressive disorder (MDD), borderline personality disorder (BPD), social anxiety disorder, obsessive-compulsive disorder (OCD), and panic symptoms (Akbari et al., 2022; Brown et al., 2022; Garner et al., 2018; Mattingley et al., 2022; Michel et al., 2016), and it is typically inversely related with psychopathology, such that a lower level of distress tolerance is associated with greater symptom severity. Distress tolerance has been examined in a range of internalizing psychopathology, but especially distress disorders (Watson, 2009), such as MDD and PTSD. Distress disorders are characterized by high negative affect and internalizing distress symptoms (Kotov et al., 2021; Watson, 2009). In addition to associations with distress disorder symptom severity, distress tolerance has been found to be negatively associated with neuroticism (Kaiser et al., 2012), negative affect intensity (Vujanovic et al., 2013), and negative affect (Jeffries et al., 2016). Thus, there are consistent empirical associations between emotional distress tolerance, distress disorders, and negative affect.
More specifically, ample research has examined the specific associations between PTSD and emotional distress tolerance. Perceived emotional distress tolerance has been negatively associated with PTSD symptom severity (Banducci et al., 2017; Vujanovic et al., 2011, 2016), as well as all four symptom clusters of DSM-5-TR PTSD (i.e., intrusions, avoidance, negative alterations in cognition and mood, and alterations in arousal and reactivity; Fetzner et al., 2014; Vujanovic et al., 2013). A recent meta-analysis found the overall effect size of PTSD symptom severity and self-reported distress tolerance to be r = −.422 (Akbari et al., 2022). Theoretically, distress intolerance and emotional avoidance are core features of PTSD symptom maintenance, as the emotional processing theory of PTSD indicates that emotional engagement is required for trauma recovery (Foa & Kozak, 1986). Vujanovic and Zegel (2020) proposed several theoretical pathways to support the consistent relationship between PTSD and distress tolerance. For instance, as a trait-like construct, low emotional distress tolerance may be a predisposing factor for experiencing more PTSD symptoms following a traumatic event, or they proposed that, following a traumatic event, emotional distress tolerance may change given the increased experience of negative affect, physiological anxiety symptoms, or re-experiencing symptoms (Vujanovic & Zegel, 2020). Likely, the relationship may have interactive or bidirectional effects when considering the role of emotional distress tolerance and PTSD symptom severity, and thus further research is warranted.
One of the most commonly used self-report measures of emotional distress tolerance is the Distress Tolerance Scale (DTS; Simons & Gaher, 2005). The DTS is a 15-item self-report measure of emotional distress tolerance, and it consists of four subscales, tolerance, absorption, appraisal, and regulation. It has been used to examine emotional distress tolerance in undergraduate student samples (Cougle et al., 2013; Glassman et al., 2016; Simons & Gaher, 2005), veteran samples (Banducci et al., 2017; Vinci et al., 2016), treatment-seeking samples (Galiano et al., 2024), and community adults (Boffa et al., 2018; Fetzner et al., 2014; Rogers et al., 2020). Further, the DTS is commonly used in trauma-exposed samples and associated with PTSD symptom severity (Byllesby & Palmieri, 2025; Fetzner et al., 2014; Simons et al., 2021; Vujanovic et al., 2011, 2013).
In the initial validation study, Simons and Gaher (2005) found, in two separate college student samples, that men reported higher overall perceived emotional distress tolerance compared to women. Other subsequent studies have found that men have higher DTS total scores compared to women (Cougle et al., 2013) or that binary gender (men vs. women) is correlated with DTS total score (Vujanovic et al., 2016). Similarly, in a clinical sample of children and adolescents, biological sex (male vs. female) was associated with DTS scores (Tonarely & Ehrenreich-May, 2020). However, not all studies have identified gender differences in DTS scores (Banducci et al., 2017; Glassman et al., 2016). In fact, in a study of community adults with a PTSD diagnosis, pre-treatment DTS score was not correlated with sex, while post-treatment DTS score was correlated with sex (Boffa et al., 2018), indicating mixed results of the relationship between the DTS and gender/biological sex.
In addition to potential gender differences, one of the other concerns regarding the DTS is its underlying factor structure. In the initial exploratory factor analysis (EFA), there was one strong eigenvalue, as well as three eigenvalues less than one; however, subsequent confirmatory factor analysis (CFA), indicated a hierarchical structure, with four lower-order correlated factors loading onto one overall distress tolerance factor (Simons & Gaher, 2005). Subsequent psychometric investigations have similarly found that the DTS is not best represented by a single, unidimensional construct (Leyro et al., 2011; Rogers et al., 2020). Leyro et al. (2011) found support for the hierarchical model with four correlated factors in a sample of community adults. Rogers et al. (2020) found that, rather than a hierarchical model, the best fitting model was a bifactor model, in which each DTS item loads on both a specific factor (e.g., Absorption) and a general overall factor (e.g., General Distress Tolerance) that is orthogonal or uncorrelated with the specific factors. In treatment-seeking adults, Galiano et al. (2024) also found support for a bifactor model, and they additionally found that only the Regulation specific factor contributed meaningful variance above and beyond the General Distress Tolerance factor. Given the uncertainty of the structure of the DTS as either primarily a single underlying construct or specific factors contributing to an overarching construct, further clarification of the construct validity of the DTS is needed.
Recently, a short form version of the DTS, the DTS-SF (Garner et al., 2018), was created. In the initial psychometric investigation, Garner et al. (2018) derived the four-item questionnaire by identifying the item from each of the four subscales with the highest factor loading in patients with OCD in residential treatment. They found good initial reliability, validity, and factor loadings for the abbreviated version. Subsequently, in a sample of undergraduate students, Brown et al. (2022) examined the DTS-SF. They found the DTS-SF scores differed by level of depression symptom severity, and the DTS-SF had good reliability, was highly correlated with the full length DTS, and good convergent validity with emotion regulation and depression. Further, when Brown et al. (2022) examined the performance of the DTS-SF by depression symptom severity, using measurement invariance testing, they found the DTS-SF had configural and metric invariance but did not meet the criteria for scalar invariance, indicating the DTS-SF has different item intercepts across depression severity groups. This implies that the DTS-SF has different means across levels or groups of depression severity, which is to be expected if there are differences in emotional distress tolerance related to more severe psychopathology. In fact, in that study, the authors reported that the best model fit for both the DTS and DTS-SF was among individuals in the highest severity group.
The present study aimed to replicate and extend the initial psychometric findings regarding the DTS-SF. Previous studies have examined the DTS-SF relative to OCD symptoms (Garner et al., 2018) and depressive symptom severity (Brown et al., 2022); however, they have not examined the relationship between the DTS-SF and PTSD symptom severity in a trauma-exposed sample. Given the robust relationship between emotional distress tolerance and PTSD symptom severity (Akbari et al., 2022), it is important to examine the relationship between the abbreviated DTS and PTSD symptoms. With this background in mind, the present study examined the associations between the DTS-SF and internalizing psychopathology (i.e., PTSD and depression symptoms) and affectivity to further extend construct validity of the DTS-SF. Also, given the mixed findings related to possible gender differences in distress tolerance, the current study intended to examine the relationship between binary gender and DTS-SF. Additionally, measurement invariance testing was used to examine if the latent factor structure of the DTS-SF fit well for two groups (exposure to traumatic event vs. no history of exposure to traumatic event, and men vs. women). Previous investigations of the DTS-SF have not examined the role of trauma exposure nor reported on gender differences in DTS-SF scores (Brown et al., 2022; Garner et al., 2018). First, it was hypothesized that there was a negative association between DTS-SF and PTSD and depression symptom severity, as well as negative affect; however, it was hypothesized there would be a positive association between the DTS-SF and positive affect. Second, it was hypothesized that those with trauma exposure would score lower on DTS-SF compared to those with no history of trauma exposure. Third, we expected to find observed score differences in the DTS-SF by gender, with males scoring higher on DTS-SF compared to females. No a priori hypotheses regarding the measurement invariance testing were set given the limited available research on the DTS-SF factor structure to date.
Method
Participants and Procedure
The sample consisted of students from two U.S. universities, one collected at a large Hispanic-serving institution in the Southwestern United States, and one collected at a regional comprehensive university in the Southeastern United States. Institutional Review Boards (IRB) from both institutions independently approved the study methodology. Sample 1 consisted of 533 undergraduate students, and Sample 2 consisted of 597 undergraduate students. These samples were combined for an overall sample of 1,130. Of these, 31 individuals were excluded because they were missing all items of the primary measure of interest (Distress Tolerance Scale). Thus, the final sample consisted of 1,099 participants (Sample 1 n = 530, Sample 2 n = 569).
Sample 1 participants were on average 20.13 years old (SD = 2.37), and most identified as female (69.8%). The majority also reported their race as White (80%), 100% reported their ethnicity as Hispanic (all other racial/ethnic groups constituted < 1% of the total sample), and most (86%) reported their sexual orientation was heterosexual. In Sample 2, participants were on average 19.83 years old (SD = 4.43), and many identified as female (60.1%; 2.7% identified as transgender or non-binary). In this sample, 80.5% identified as White, 11.4% as Black/African American, 8.8% as Hispanic, and 1.9% as American Indian/Indigenous, and participants were able to indicate multiple race and ethnicity categories as appropriate. For sexual orientation, 76.4% identified as heterosexual and 14.1% identified as bisexual. Regarding trauma exposure, 214 (40.2%) participants in Sample 1 reported experiencing a potentially traumatic event, and 328 (54.9%) reported experiencing a potentially traumatic event in Sample 2, resulting in 542 (48.0%) trauma-exposed participants overall.
Instrumentation
Distress Tolerance Scale (DTS; Simons & Gaher, 2005). The DTS is a 15-item self-report measure of emotional distress tolerance. Items are rated on a five-point scale, where 1 is “Strongly agree” and 5 is “Strongly disagree,” and one item is reverse coded. Higher scores indicate better tolerance of distress. The DTS is often examined with four subscales: absorption, tolerance, appraisal, and regulation. Absorption (three items) measures the attention being absorbed by the presence of negative emotional states, and tolerance (three items) is specifically the perceived ability to tolerate or withstand one’s emotional distress. Appraisal (six items) assesses the subjective appraisal of distress, and regulation (three items) considers the efforts to regulate or alleviate distress. Previous studies have found good test-retest reliability and convergent and divergent validity in college samples and clinical samples (Galiano et al., 2024; Simons & Gaher, 2005). Reliability was good in the overall sample, with Cronbach’s α = 0.93 (trauma α = 0.93, no trauma α = 0.92, male α = 0.93, female α = 0.92).
Distress Tolerance Scale-Short Form (DTS-SF; Garner et al., 2018). The DTS-SF is a four-item self-report measure of emotional distress tolerance that was derived from the 15-item DTS. The four items were included by selecting the item with the highest factor loading from each of the four original subscales (My feelings of distress are so intense that they completely take over. Being distressed or upset is always a major ordeal for me. I can’t handle feeling distressed or upset. I’ll do anything to stop feeling distressed or upset.). Items are rated on the same five-point scale (1 = Strongly agree and 5 = Strongly disagree), with higher scores associated with more tolerance of distress. Participants completed the DTS-SF items as they were embedded in the full length DTS. Previous research indicates the scale is unidimensional (Brown et al., 2022). Cronbach’s α = 0.83 in the overall sample (trauma α = 0.83, no trauma α = 0.83, male α = 0.84, female α = 0.83).
Stressful Life Events Screening Questionnaire (SLESQ; Goodman et al., 1998). Sample 1 used the SLESQ to screen for exposure to potentially traumatic events. The SLESQ assesses lifetime prevalence to 12 potentially traumatic events, as well as providing an “other” option for participants. The SLESQ was modified to include an option for participants to indicate their most distressing traumatic event. The most commonly reported index events were family member or close friend dying by accident, homicide, or suicide (35.0%), childhood unwanted sexual experience (16.8%), and life-threatening accident (7.5%).
Life Events Checklist for DSM-5 (LEC-5; Weathers et al., 2013). Sample 2 used the LEC-5 to screen for lifetime exposure to potentially traumatic events. The LEC-5 describes 16 potentially traumatic events and participants are asked to indicate if they experienced any of these, with the options “happened to me,” “witnessed it,” “learned about it,” “part of my job,” “not sure,” or “doesn’t apply.” It also includes an “other” category for events that are not otherwise captured. Participants were then asked to indicate their most distressing traumatic event. The most commonly reported index events were life-threatening accident (24.1%), sexual assault (19.2%), and family member or close friend dying by accident, homicide, or suicide (14.0%).
PTSD Checklist for DSM-5 (PCL-5; Weathers et al., 2013). The PCL-5 is a 20-item self-report measure of PTSD symptom severity, where each item is based on one symptom of the DSM-5 diagnostic criteria for PTSD. Items are rated on a five-point scale (0 = Not at all and 4 = Extremely) of how bothersome they find each symptom within the past month. Only participants who indicated they had experienced a Criterion A potentially traumatic event on the SLESQ or LEC-5 were given the PCL-5. Participants were asked to rate each item anchored to their most distressing potentially traumatic event, as assessed on either the SLESQ or LEC-5. The PCL-5 has demonstrated good internal consistency, test-retest reliability, convergent and discriminant validity, and predictive validity in previous studies (Blevins et al., 2015; Bovin et al., 2016; Lee et al., 2019). The scale had excellent reliability in the present sample, Cronbach’s α = 0.96.
Patient Health Questionnaire-9 (PHQ-9; Kroenke et al., 2001). The PHQ-9 is a nine-item self-report measure of depression symptom severity, based on the nine symptoms of major depressive disorder in the DSM-5. The PHQ-9 has participants rank the frequency of their depressive symptoms over the past two weeks on a four-point scale (0 = Not at all to 3 = Nearly every day). The PHQ-9 has demonstrated good reliability, criterion validity, and convergent and divergent validity (Beard et al., 2016; Kroenke et al., 2001). In the current sample, the scale had excellent internal reliability, Cronbach’s α = 0.94.
Positive and Negative Affect Scale (PANAS; Watson et al., 1988). The PANAS is a 20-item self-report measure of positive affect (10 items) and negative affect (10 items). Participants are asked to rate their affect in the past month on a five-point scale (1 = Very slightly or not at all to 5 = Extremely). The PANAS has demonstrated good test-retest reliability, internal consistency, and validity (Watson et al., 1988). The PANAS had good internal reliability in the present sample, negative affect Cronbach’s α = 0.91 and positive affect Cronbach’s α = 0.90.
Data Analyses
Initial data screening procedures and descriptive statistics were conducted using SPSS version 28. First, data were screened for missingness, and participants with more than 20% missing item-level data on the DTS were excluded. As stated above, 31 individuals were missing the entire DTS, resulting in 1099 overall participants. An additional 27 participants were missing one or two items on the DTS, 17 were missing 1–2 items on the PANAS, nine were missing a single item on the PHQ-9, and 13 were missing 1–2 items on the PCL-5. Item-level data were estimated using maximum likelihood (ML) estimation with a pairwise present approach in Mplus (see below). Total scores were calculated and screened for normality. The DTS-SF was examined for convergent validity with the full length DTS. Further validity was explored based on sample demographics, measures of psychopathology (i.e., PCL-5, PHQ-9) and affectivity (i.e., PANAS), and latent factor structure.
Additional analyses were conducted in Mplus Version 8 software (Muthén & Muthén, 2017). Several confirmatory factor analyses (CFA) were conducted on the DTS and DTS-SF to determine model fit using the overall sample (n = 1099). Items on both versions of the DTS were treated as ordinal because they have only five response options (Wirth & Edwards, 2007), so the CFAs used robust weighted least squares estimation with a mean- and variance-adjusted chi-square (WLSMV) estimation, a polychoric covariance matrix, and probit regression coefficients. Factor variances were fixed to one. Model fit was interpreted using Hu and Bentler (1999) criteria, with excellent model fit being indicated by a comparative fit index (CFI) and Tucker-Lewis Index (TLI) ≥ 0.95 and root mean square error of approximation (RMSEA) values < 0.06. Standardized model estimates were interpreted. First, the fit for the unidimensional four-item DTS-SF model was examined. As a comparison, three models of the DTS were examined to determine best model fit: four correlated factors, four correlated factors with a higher order factor, and four correlated factors with a bifactor. Given previous research (Leyro et al., 2011; Rogers et al., 2020), we expected either the hierarchical or bifactor model to provide better fit.
Finally, to extend previous investigations of the DTS-SF, we examined the DTS-SF unidimensional factor structure for measurement invariance, comparing model fit between individuals with a reported history of trauma and those without a history of trauma, to determine if the DTS-SF factor structure demonstrates invariance. Broadly speaking, measurement invariance tests if the scale is measuring the same thing in the same way across different groups. Therefore, measurement invariance was tested based on reported gender (male vs. female). Models tested configural invariance (Model A), metric invariance (Model B), and finally scalar invariance (Model C) across the two groups, and models were examined using maximum likelihood estimation with robust standard errors (MLR) because WLSMV does not allow for measurement invariance testing. Configural invariance is the least restrictive model, and configural invariance indicates the DTS-SF has the same basic organization of the construct across groups (Putnick & Bornstein, 2016). Next, metric invariance was tested by constraining factor loadings to be equal between groups. Then, item intercepts and factor loadings were constrained to be equal across groups to test scalar invariance. Models were examined by assessing the change in CFI values, as suggested by Cheung and Rensvold (2002), as ΔCFI is less dependent on sample size and model parameters compared to the chi-square statistic. A CFI difference between two models (e.g., configural vs. metric) of < 0.01 indicates that the models are not invariant, and the null hypothesis should not be rejected.
Results
Means, standard deviations, range, and correlations are presented in Table 1. Of those with a history of trauma exposure, 32.2% (n = 172) had a PCL-5 total score greater than or equal to 33 (Bovin et al., 2016), indicating probable PTSD diagnosis. On the PHQ-9, 36.9% of the sample (n = 406), had a PHQ-9 score of 10 or greater, indicating probable major depressive disorder (Kroenke et al., 2001). The DTS (r =.090, p =.005) and the DTS-SF (r =.077, p =.017) had a low correlation with age in the present sample, thus older participants reported better perceived emotional distress tolerance, but the effect was small. When examining gender differences, individuals identifying as non-binary or transgender were not included. Participants who identified as male (M = 14.16, SD = 4.02) had higher DTS-SF scores compared to females (M = 12.73, SD = 4.14), t(1081) = 5.46, p <.001, Cohen’s d = 0.349, 95% confidence interval (CI) [0.223, 0.476], indicating a small effect. The effect was similar using the DTS, with males (M = 50.69, SD = 12.84) reporting higher DTS scores compared to females (M = 46.49, SD = 13.10), t(1081) = 5.05, p <.001, Cohen’s d = 0.323, 95% CI [0.197, 0.449]. There was also a significant difference in DTS-SF total score for individuals reporting a history of trauma (n = 534, M = 12.82, SD = 4.19) and those without a potentially traumatic event (n = 565, M = 13.54, SD = 4.10), t(1097) = 2.89, p =.004, Cohen’s d = 0.175, 95% CI [0.056, 0.293], such that individuals with a history of trauma reported lower distress tolerance, although the effect size was small. Similarly, there was a small difference in the DTS total score when comparing individuals with a history of trauma (M = 46.74, SD = 13.33) and individuals without a history of trauma exposure (M = 48.80, SD = 12.96), t(1097) = 2.60, p =.010, Cohen’s d = 0.157, 95% CI [0.038, 0.275], and this was also a small effect size.
Table 1
Means, standard Deviations, and bivariate correlations
Mean (SD) | Obs Range | DTS | DTS-SF | PCL-5 | PHQ-9 | PANAS-N | |
|---|---|---|---|---|---|---|---|
DTS | 47.79 (13.18) | 16–75 | - | ||||
DTS-SF | 13.19 (4.16) | 4–20 | 0.954* | - | |||
PCL-5 | 24.48 (20.43) | 0–80 | − 0.413* | − 0.409* | - | ||
PHQ-9 | 12.72 (8.12) | 0–27 | − 0.393* | − 0.384* | 0.601* | - | |
PANAS-N | 22.82 (8.79) | 10–50 | − 0.398* | − 0.385* | 0.605* | 0.633* | - |
PANAS-P | 31.14 (9.31) | 10–50 | 0.207* | 0.206* | − 0.140* | − 0.300* | 0.011 |
The relationships between the DTS and DTS-SF with psychopathology and affect are also presented in Table 1. The DTS-SF demonstrated similar strengths of relationships when compared to the full length DTS. As anticipated, the DTS-SF was negatively associated with PTSD symptom severity (r = −.409), depression symptom severity (r = −.384), and negative affect (r = −.385), but the DTS-SF was positively associated with positive affect (r =.206).
Next, the factor structure of the DTS and DTS-SF were examined. The unidimensional DTS-SF model had adequate model fit, χ2(2) = 50.56, p <.001, CFI = 0.991, TLI = 0.973, RMSEA = 0.148, 90% CI [0.115, 0.185]. Factor loadings (see Table 2) ranged from 0.635 to 0.875, and there were no modification indices that suggested an improvement in model fit. This was then compared to several latent models of the DTS (see Table 2 for factor loadings). The model consisting of four correlated factors had adequate fit, χ2(84) = 1115.88, p <.001, CFI = 0.959, TLI = 0.949, RMSEA = 0.106, 90% CI [0.100, 0.111]. The four-factor model that included a higher order factor had similar model fit, χ2(86) = 1249.67, p <.001, CFI = 0.954, TLI = 0.944, RMSEA = 0.111, 90% CI [0.106, 0.116]. Model fit for the DTS improved slightly in the model examining four correlated factors with an orthogonal general factor (i.e., the bifactor model), χ2(69) = 620.40, p <.001, CFI = 0.978, TLI = 0.967, RMSEA = 0.085, 90% CI [0.079, 0.091]. For the General Distress Factor, ω = 0.948, indicating high internal consistency, and the ωH = 0.793. For the other factors, ωHS were tolerance = 0.461, appraisal = 0.383, absorption = 0.401, and regulation = 0.229.
Table 2
Factor loadings for initial confirmatory factor analyses
Item | Hierarchical | Correlated | Bifactor model | Short form | |||
|---|---|---|---|---|---|---|---|
model | factors model | Specific factor | General factor | ||||
Tolerance | |||||||
DTS1 | 0.810 | 0.810 | 0.611 | 0.538 | |||
DTS3 | 0.868 | 0.867 | 0.707 | 0.542 | 0.845 | ||
DTS5 | 0.778 | 0.779 | 0.477 | 0.594 | |||
Appraisal | |||||||
DTS6 | − 0.020* | − 0.023* | 0.369 | − 0.250 | |||
DTS7 | 0.686 | 0.686 | 0.182 | 0.693 | |||
DTS9 | 0.777 | 0.776 | 0.511 | 0.595 | |||
DTS10 | 0.864 | 0.865 | 0.594 | 0.647 | 0.808 | ||
DTS11 | 0.829 | 0.829 | 0.485 | 0.672 | |||
DTS12 | 0.844 | 0.844 | 0.539 | 0.657 | |||
Absorption | |||||||
DTS2 | 0.837 | 0.836 | 0.639 | 0.570 | |||
DTS4 | 0.877 | 0.875 | 0.651 | 0.609 | 0.875 | ||
DTS15 | 0.858 | 0.861 | 0.430 | 0.741 | |||
Regulation | |||||||
DTS8 | 0.821 | 0.821 | 0.392 | 0.706 | |||
DTS13 | 0.923 | 0.923 | 0.592 | 0.767 | 0.635 | ||
DTS14 | 0.772 | 0.773 | 0.296 | 0.692 | |||
Total | |||||||
Tolerance | 0.947 | ||||||
Appraisal | 0.935 | ||||||
Absorption | 0.977 | ||||||
Regulation | 0.749 | ||||||
Measurement invariance model fit indices for the DTS-SF are provided in Table 3. Regarding trauma exposure, there was no change in CFI value comparing the configural and metric model, indicating that the groups have the same underlying factor structure and invariant factor loadings. In contrast, the ΔCFI = 0.016 comparing the metric and scalar models, indicating non-invariance on item intercepts when comparing individuals with a history of trauma exposure to those without a history of trauma exposure.
Table 3
Measurement invariance testing
Model | χ2 (df) | p | CFI | TLI | RMSEA [90% CI] | BIC | ΔCFI |
|---|---|---|---|---|---|---|---|
Measurement invariance results for trauma exposure groups | |||||||
Configural | 3.188 (2) | 0.203 | 0.999 | 0.994 | 0.033 [0.000, 0.097] | 12946.809 | |
Metric | 5.869 (5) | 0.319 | 0.999 | 0.998 | 0.018 [0.000, 0.064] | 12928.142 | 0.000 |
Scalar | 28.603 (9) | 0.001 | 0.983 | 0.977 | 0.063 [0.038, 0.090] | 12924.031 | 0.016 |
Measurement invariance results for gender | |||||||
Configural | 2.697 (2) | 0.260 | 0.999 | 0.996 | 0.025 [0.000, 0.093] | 12713.213 | |
Metric | 8.280 (5) | 0.142 | 0.997 | 0.993 | 0.035 [0.000, 0.075] | 12698.167 | 0.002 |
Scalar | 43.432 (9) | < 0.001 | 0.968 | 0.957 | 0.084 [0.060, 0.110] | 12708.026 | 0.029 |
Measurement invariance was also tested for binary gender for the DTS-SF, and the results are provided in Table 3. For these analyses, only 1083 participants (n = 371 males and n = 712 females) were used in the analyses, and individuals who identified as non-binary or transgender were excluded from the group comparison. There was no change in CFI value comparing the configural and metric model, indicating that males and females have the same underlying factor structure and invariant factor loadings on the DTS-SF. However, the ΔCFI = 0.029 comparing the metric and scalar models, indicating non-invariance on item intercepts between males and females.
Discussion
The current study aimed to investigate the psychometric properties of the Distress Tolerance Scale Short Form (Garner et al., 2018), extending recent research on the abbreviated version of the DTS. Consistent with prior studies (Brown et al., 2022; Garner et al., 2018), the DTS-SF was highly correlated with the full length DTS. Although the DTS-SF did have lower internal reliability relative to the DTS, this is to be expected of a shorter measure (Schmitt, 1996). In support of hypothesis 1, the DTS-SF demonstrated expected patterns of convergent validity (negative correlations) with measures of PTSD symptom severity, depression symptom severity, and negative affect, as well as a positive association with positive affect. It also appeared that these correlations were of a similar magnitude when compared to the full length DTS, further supporting the use of the DTS-SF as a measure of emotional distress tolerance. Male participants reported higher scores on the DTS-SF than females (hypothesis 2 supported), and those with exposure to trauma (vs. no trauma exposure) reported lower DTS-SF scores (hypothesis 3 supported), although the effect sizes were small. These findings along with those of the measurement invariance are detailed below.
The factor structure of the DTS-SF had generally good model fit. The initial CFA using the entire sample had high a RMSEA value, which was consistent with Brown et al. (2022). It is worth noting that RMSEA performs worse (i.e., has higher values) when there are fewer degrees of freedom in the model (Kenny et al., 2014). Given the unidimensional structure of the DTS-SF, and that it only has two degrees of freedom in the CFA, it might be expected to have an elevated RMSEA value. The incremental fit indies (CFI and TLI) supported that the unidimensional structure fit well relative to the baseline model. Concerning the full length DTS, it appeared the absolute and incremental fit indices favored the bifactor model, which is consistent with previous studies (Galiano et al., 2024; Rogers et al., 2020). The appeal of a bifactor model is that it separates out the generalized variance (i.e., the general factor) from the variance associated with the specific factors to examine the dimensionality of the scale (Reise, 2012). The general factor had high reliability and explained much of the common variance; however, there was also variance accounted for by the specific factors above and beyond the general factor, indicating multidimensionality. Specifically, the tolerance factor explained the most variance above and beyond the general distress tolerance factor, and, contrary to Rogers et al. (2020), the regulation factor had the lowest omega hierarchical subscale score. The omega hierarchical subscale score represents the unique variance of each subscale score after accounting for the variance of the general factor (Reise et al., 2013).
The present study is the first study to examine the performance of the DTS-SF in an explicitly trauma-exposed sample, though other studies likely had some participants with a history of trauma exposure. At the observed total score level, the DTS-SF and the full length DTS total scores differed by trauma exposure, such that individuals with a history of PTSD Criterion A trauma exposure reported lower tolerance of emotional distress compared to individuals who did not report a history of trauma. As noted above, the DTS-SF was also negatively associated with PTSD symptom severity, as hypothesized. This is consistent with previous research that has found a robust relationship between PTSD symptom severity and emotional distress tolerance (Akbari et al., 2022). Considering the measurement invariance results, metric invariance was supported, indicating equivalence of factor loadings in the two groups. Thus, all four items have factor loadings that specify the underlying factor in a similar way and at approximately the same strength or magnitude across the two groups (Putnick & Bornstein, 2016). In other words, achieving metric invariance indicated that those with trauma exposure and those with no trauma exposure, interpreted the DTS-SF scale in a similar way, allowing for meaningful comparisons of the latent variable of DTS-SF. Scalar, or strong factorial, invariance was not supported, so there were group mean differences (e.g., non-invariance) of item intercepts when comparing participants with and without a history of a Criterion A trauma. Thus, there were differences in the mean item responses on the scales across the groups (Marsh et al., 2018). Findings suggest that only latent level mean group differences on the DTS-SF scores should be employed after achieving metric invariance, as groups (trauma exposed vs. not) interpret the scale differently. This further supports the group differences found in the total scores, though it does not indicate latent level mean differences. This is somewhat consistent with the results of Brown et al. (2022), who also found that the DTS-SF did not meet the criteria for scalar invariance when they examined the measurement invariance of the DTS-SF based on depression symptom severity. These results support the use of the DTS-SF in trauma-exposed samples, while also identifying that there may be measurement differences in participants with a history of trauma exposure compared to those without trauma-exposure.
When examining measurement invariance of DTS-SF across males and females, the observed total score means differed such that men reported higher tolerance for emotional distress compared to women, on both the DTS and DTS-SF, though the effect was small. This provides additional support to other studies that have identified differences in gender or sex on the DTS (Simons & Gaher, 2005; Tonarely & Ehrenreich-May, 2020; Vujanovic et al., 2016) and extends it to the DTS-SF. Subsequent analyses at the latent factor level, assessed using measurement invariance, indicated invariance of factor structure and factor loadings; however, men and women did differ in their item intercepts on the DTS-SF. This is consistent with there being group differences in the item interpretations of emotional distress tolerance on DTS-SF, with men reporting higher levels of overall emotional distress tolerance relative to women, as well as high item intercepts.
The present study cannot provide support for gender differences in the construct of emotional distress tolerance itself, only the possibility of differential performance of one measure used to assess the construct. As emotional distress tolerance is conceptualized as a trait-like construct, there are no theoretical reasons to expect gendered differences in the underlying latent construct. One potential confound in this association is the relationship between gender identity and biological sex with psychopathology, such that there may be differential severity or prevalence rates by these demographic constructs (Hartung & Lefler, 2019). For example, previous research has found that women report higher rates of PTSD symptom severity (e.g., Ditlevsen & Elklit, 2012), so women may be experiencing higher levels of psychopathology and overall distress, which may be influencing their reported self-efficacy in managing that distress. In post hoc analyses examining the gender differences in our convergent validity variables (e.g., the PCL-5, PHQ-9, and PANAS), we did find that women reported higher levels of psychopathology and negative affect relative to men. Given that the present study examined exclusively internalizing psychopathology (e.g., PTSD and depression) and its relationship to distress tolerance, and that women report higher means of internalizing psychopathology and men report higher means on externalizing psychopathology (Eaton et al., 2012), it may be confounding the results related to gender identity and distress tolerance. Additional research would be beneficial to better understand the relationship between emotional distress tolerance and binary gender.
Given the summary of the present findings, it appears that DTS-SF may be appropriate to use as an abbreviated form of the DTS to measure self-reported emotional distress tolerance. It demonstrates good internal reliability (both Cronbach’s alpha and omega), and is strongly correlated with the full-length DTS, indicating good construct validity. It demonstrated the anticipated relationships with psychopathology, such that it was inversely related to symptom severity of PTSD and depression, and positively associated with positive affect, supporting convergent validity. Although it did not demonstrate scalar invariance in the present study, scalar invariance is a high threshold for many measures to attain (Marsh et al., 2014), and future research may help clarify the circumstances, contexts, or populations where its use is most appropriate. There may be some overlap in DTS-SF item content and PTSD avoidance symptoms as described in the PCL-5, such that item similarity may contribute to the associations between measures. Future research should consider teasing apart distress intolerance, PTSD avoidance symptoms, and experiential avoidance. Additional research in samples with greater symptom severity may further assist understanding these relationships. For example, individuals with a history of interpersonal vs. non-interpersonal trauma may have differential ratings of their distress tolerance, or distress tolerance and its measurement may be moderated by level of symptom severity (e.g., low/no symptoms, subthreshold PTSD symptoms, vs. meeting the diagnostic criteria for PTSD).
The present results should be considered within the context of some limitations. Although the sample size was large, it was a convenience sample of college students, and thus it may not generalize well outside of this sample. The data were also collected exclusively through self-report, which may impact external validity and there is a possibility of response biases. In contrast to the initial development and validation of the DTS-SF, the present sample consisted of non-treatment seeking participants with lower levels of psychopathology severity. Given the use of self-report and anonymous data, it cannot be known if all participants had a clearly defined Criterion A trauma that they used as an anchor for their PTSD symptom severity rankings on the PCL-5. Despite this, we used a large overall sample to examine the psychometric properties of this scale thus were adequately powered (Meade & Bauer, 2007). The present study expanded the understanding of the measurement performance, reliability, and validity of the DTS-SF, which is a newly developed self-report measure that may be appropriate to use when time or participant burden are concerns for researchers.
Declarations
Competing Interests
The authors report there are no competing interests to declare.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
