Authors: Sonalee A. Joshi, D. Angus Clark, James L. Abelson, Elizabeth R. Duval
Categories: Original Research, Negative urgency, Anxiety, Impulsivity, Hyperactivity, Experiential avoidance
Source: Journal of Mood and Anxiety Disorders
Authors: Sonalee A. Joshi, D. Angus Clark, James L. Abelson, Elizabeth R. Duval
Anxiety and impulsivity often co-occur, but due to broad conceptualizations of impulsive behavior, it remains unclear whether anxiety could be positively associated with global impulsivity or specific facets of impulsivity. Negative urgency, or rash behavior that specifically occurs during negative emotional states, may provide common ground to examine the overlap between high anxiety and high impulsivity. Limited work has examined relationships between negative urgency and behavioral processes that are often associated with both anxiety and global impulsivity, such as experiential avoidance. In this study, we examined self-report data collected from a non-clinical sample of adult participants (n = 589) to test relationships between anxiety and experiential avoidance through both negative urgency and global impulsivity. Contrary to hypotheses, we found that negative urgency alone did not alter the relationship between anxiety and experiential avoidance. However, follow-up analyses including data-driven factors of impulsivity revealed that in combination, anxiety and both emotion-based impulsivity (i.e. negative and positive urgency) and hyperactivity were linked to greater experiential avoidance. By contrast, anxiety and non-planning were associated with lower experiential avoidance. These results provide targets for future studies to examine individual differences in profiles of co-occurring anxiety and impulsivity.
Anxiety disorders are highly prevalent (7.3%; [1] and are often marked by clinically significant overregulation and avoidance [2]. By contrast, impulsivity is broadly defined by difficulty inhibiting behavioral urges and increased approach-based motivational processes [3] and is emblematic of externalizing disorders (e.g. Attention Deficit Hyperactivity Disorder [ADHD], and substance use disorders [SUD]; [4], [5], [6]). Historically, trait impulsivity was considered the opposite of anxious avoidance, and these conditions were studied in isolation from each other, using separate metrics not sensitive to their potential intersections [7]. However, anxiety and impulsivity frequently co-occur, and there are strong comorbidities between anxiety and externalizing disorders commonly linked to high impulsivity. Approximately 25% of people with ADHD and up to 21% of people with substance use disorders also meet criteria for an anxiety disorder [8], [9], [10].
More recent transdiagnostic approaches have attempted to capture the varied forms in which anxious and impulsive symptomatology may present together [11], [12]. For example, dangerous behaviors can be undertaken to escape a fear-inducing situation (e.g., an individual with a driving phobia speeding to minimize time on the highway). Alternatively, impulsive reward-driven behavior (like substance use), can also be undertaken to escape the internal distress generated by fear and anxiety (e.g., an individual with social anxiety binge drinking to cope with discomfort while socializing) [13]. In both cases, impulsive, risky behavior is connected to anxiety-driven avoidance.
Closely examining the role that impulsivity plays in inciting avoidance behavior may help us to better understand links between impulsivity and anxiety. Limitations of prior studies include the use of overly broad definitions of impulsivity, inconsistent examination of only some dimensions of impulsivity, and discrepancies across measurement methods (i.e., behavioral indices vs. self-report measures). These limitations have complicated research examining the nuanced relationship between anxiety and impulsivity [14], [15], [16]. Non-emotional dimensions of impulsivity, such as reduced motor inhibition [17], [18], and hyperactivity [19], [20], have been positively associated with anxiety, suggesting links between elevated anxiety and heightened impulsive behavior. However, emotion-based dimensions of impulsivity may have particular relevance to internalizing disorders, including anxiety. Prior research has posited various forms of psychopathology stem from a combination of high emotional reactivity and behavioral disinhibition [21]. Negative urgency is an emotion-based dimension of impulsivity defined by rash behavior brought on by negative emotion [22]. Thus, negative urgency captures the combination of emotional distress and impulsive behaviors to reduce distress and has relevance to both internalizing and externalizing disorders [23], [24], [25], [26], [27]. Therefore, negative urgency is an impulsive dimension that is well positioned to examine links between anxiety and impulsivity marked by distress intolerance and reactive avoidance.
Experiential avoidance is defined by intolerance of emotional and physiological pain or distress and often manifests as short-term reactionary efforts (e.g., substance use, suppression of negative thoughts) to alleviate or avoid this distress [28]. Experiential avoidance has been linked to several anxiety disorders [29], [30], [31], [32], greater externalizing disorder symptom severity (e.g., SUD; [33], ADHD; [34]), and greater engagement in risky behaviors (e.g., sexual behaviors [35], binge eating; [36]), non-suicidal self-injury (NSSI; [37]). Despite relationships with externalizing pathology, only one study has examined experiential avoidance and negative urgency together. This study reported a positive correlation between experiential avoidance and negative urgency in a sample of individuals with disordered eating [38]. Importantly, these relationships parallel those documented between similar short term, reflexive coping behaviors (e.g., distraction, avoidance, suppression) and negative urgency [39].
Although prior studies have examined relationships between impulsivity and anxiety, few studies have specifically examined which dimensions of impulsivity may have the strongest associations with anxious avoidance. In this study, we aimed to characterize relationships between anxiety, dimensions of impulsivity, and experiential avoidance using structural equation modeling (SEM). Due to prior literature linking emotion-based impulsivity to anxiety, we tested two specific 1) that anxiety, negative urgency, global impulsivity, and experiential avoidance would all be positively related, and 2) that there would be a stronger positive effect of negative urgency (compared to other dimensions of impulsivity – referred to subsequently as “global impulsivity”) on the relationship between trait anxiety and experiential avoidance. To test these hypotheses, we first conducted an exploratory factor analysis to determine the optimal factor structure of global impulsivity. Then, we conducted the confirmatory factor analysis to test the specific hypotheses that 1) there would be positive relationships between anxiety, negative urgency, global impulsivity, and experiential avoidance; and 2) that negative urgency, in combination with anxiety, would show a stronger positive relationship with experiential avoidance than global impulsivity.
The use of a well-known internet-based crowdsourcing platform, Amazon MTurk, allowed for access to a large, nationwide sample. Inclusion criteria for the study included adults between the ages of 18–55 based in the United States. In order to account for the possible impacts of the COVID-19 pandemic on self-reported anxiety and impulsivity, data were collected from three cohorts staggered across phases of the pandemic (May 2020, June 2021, and October 2021). The Fear of Illness and Virus Evaluation (FIVE; Ehrenreich-May, 2020) was administered as a measure of virus-related anxiety, but its inclusion in analyses did not improve model fit and the cohorts did not significantly differ on self-report indicators of anxiety, experiential avoidance, or impulsivity, all p’s > 0.05. Given intentions to establish proof of concept for relationships between anxiety, experiential avoidance, and negative urgency, it was important to first capture these relationships across the full spectrum of scores in an adult sample. Thus, participants were not specifically selected based on these dimensions. The final sample consisted of 589 participants (Mage= 32.05, SDage= 7.19; 79.78% White; 59.34% Male; 15.45% Sexual Minority; 43.11% < 10,000162.73%29,99910017.04%49,99913723.34%69,99912220.78%89,999447.50%149,999518.69%≥ $150000233.92%
Participants were paid consistent with United States federal minimum wage directly through the MTurk system. Information about the study was provided to all participants, who indicated their consent to participate prior to data collection. This study received an exempt status by the Institutional Review Board at the University of Michigan.
Data collection via platforms such as MTurk has become more commonplace due to benefits of more rapid and widespread sampling and greater control over research design than use of secondary datasets. However, limitations of MTurk samples have been well documented [40]. Therefore, several measures were taken to ensure data quality, consistent with prior literature (See Appendix 1; [41], [42]).
Participants were instructed to complete a series of behavioral surveys on self-reported symptoms of anxiety, trait impulsivity, and experiential avoidance.
The BIS-11 [43] is a measure of global impulsivity that consists of thirty statements regarding impulsivity that participants responded to on a Likert-type scale. Ten items were reverse scored, and total sum scores were calculated as sums of all items as a primary measure of global impulsivity. The BIS-11 has been previously demonstrated to have high internal consistency [44] and high criterion-related validity across clinical and non-clinical samples, but suboptimal construct validity [45]. Within the present sample, internal consistency for the BIS-11 total score was good with Cronbach’s alpha (α) = 0.88.
The UPPS-P [46] is a well validated measure of dimensional impulsivity that probes five dimensions of impulsivity (Negative Urgency, Lack of Premeditation, Lack of Perseverance, Sensation Seeking, and Positive Urgency). Items for each subscale are summed separately. For the current sample, internal consistency ranged from α = 0.86 (Lack of Premeditation) to α = 0.96 (Positive Urgency).
The STAI [47] includes two subscales of trait (20 items) and state (20 items) anxiety that participants responded to on a Likert-type scale. Items for each subscale were summed separately. Both trait and state subscales of the STAI have previously demonstrated high internal consistency for adult participants [47]. The STAI is one of the most widely used measures of both state and trait level anxiety [48]. This measure has been well-validated [49]. In the present sample, internal consistency for the trait subscale was α = 0.83.
The AAQ-II [35] is a seven-item measure of experiential avoidance that participants responded to using a Likert-type scale. This measure defines experiential avoidance as painful memories and experiences and consequences of emotion. The AAQ-II has been shown to have high criterion validity but suboptimal convergent and discriminant validity [50]. The AAQ-II has previously demonstrated high internal consistency [51], [52]; in the present sample, internal consistency was α = 0.94.
Structural Equation Modeling (SEM) was used to examine cross sectional relationships between anxiety, negative urgency, the latent variable of global impulsivity, and experiential avoidance. We used a series of primary hypothesis-driven and secondary exploratory models. We first conducted an EFA on subscales of the BIS-11 and UPPS-P (excluding negative urgency) using an oblique geomin rotation to inform the latent factor structure of global impulsivity in the primary SEM. The primary SEM was fit using full information maximum likelihood estimation, and the 95% confidence intervals for parameter estimates were computed using nonparametric percentile bootstrapping with 10,000 random draws. In the primary SEM, relationships were specified between trait anxiety (i.e., independent variable) to experiential avoidance (i.e., dependent variable; Fig. 1). Relationships were also specified between trait anxiety to intermediary variables of a) UPPS-P negative urgency and b) a latent factor of global impulsivity, and from these intermediary variables to experiential avoidance to test for indirect effects. In these models, global impulsivity was incorporated as a latent variable indicated by BIS-11 total score and the UPPS-P subscales, excluding negative urgency.Fig. 1Conceptual model of primary structural equation model.Fig. 1
Descriptive statistics for all variables were computed in MPlus Version 8.4 [53] and are provided in Table 2. Sum scores were normally distributed and ranged from subclinical to clinical levels. All data were standardized (Z scores computed) to account for differences in scales of the psychometric measures. Intercorrelations between measures of impulsivity included in the EFA ranged from r = −0.05 to r = 0.78 (Table 3).Table 2Descriptive Statistics.Table 2MeasureMeanSDMinMaxKurtosisSkewαUPPS-P1. Negative Urgency26.628.8412.0047.00-0.710.360.922. Positive Urgency27.5311.4714.0047.00-0.410.770.963. Sensation Seeking28.238.8412.0048.00-0.810.100.914. Lack of Premeditation19.625.7411.0044.002.331.170.865. Lack of Perseverance19.495.9310.0038.00-0.160.470.87BIS-116. Total60.1212.8134.00105.00-0.130.420.88AAQ-II7. Total23.6111.467.0049.00-1.080.320.94STAI8. State40.6910.7126.0074.00-0.200.660.899. Trait45.569.1729.0071.00-0.630.260.83Table 3Zero-order Correlation Matrix.Table 3Measure12345678UPPS-P1. Negative Urgency-2. Positive Urgency0.78-3. Sensation Seeking0.350.52-4. Lack of Premeditation0.350.340.19-5. Lack of Perseverance0.470.31-0.050.46-BIS-116. Total0.710.640.270.490.58-AAQ-II7. Total0.600.450.080.110.460.57-STAI8. State0.480.31-0.030.150.460.500.70-9. Trait0.390.460.190.140.160.410.490.40
In order to conduct the hypothesis-driven CFA, we first had to conduct an EFA using an oblique geomin rotation to inform the factor structure of global impulsivity. The EFA revealed that positive urgency (0.60) and sensation seeking (0.71) loaded strongly on one factor (i.e., greater than 0.35; [54]); while lack of perseverance (0.83), lack of premeditation (0.59), and the total BIS-11 score (0.79) loaded strongly on a second factor (Table 4). According to established standards of good fit [55], [56], this model did not meet all standards of good fit (ꭓ^2^ = 16.84, df = 1, p < 0.001; RMSEA = 0.16; SRMR = 0.02; CFI = 0.88; TLI = 0.83). However, this model was the best fitting model. Based on a review of modification indices and alternative specifications, it appeared that misfit was mainly due to cross-loadings between factors and large residual correlations that would be expected for multiple dimensions of impulsivity. We did not include all cross-loadings in our model to avoid overfitting the data. Therefore, for the latent global impulsivity factor in the primary SEM, covariances were specified between a) positive urgency and sensation seeking, and b) lack of premeditation, lack of perseveration, and the BIS-11 total score.Table 4Factor loadings from exploratory factor analysis of impulsivity subscales.Table 4EFAFactor 1Factor 2Positive Urgency0.60 *0.45 *Sensation Seeking0.71 *-0.002Lack of Premeditation0.0020.59 *Lack of Perseverance-0.31 *0.83 *BIS-110.17 *0.79 *Note. EFA = exploratory factor analysis. Oblique, geomin rotation used in EFA; factor correlation r = 0.30.
To test the hypothesis that when anxiety is high, negative urgency would be more strongly related to experiential avoidance than global impulsivity, we ran a CFA that included trait anxiety, negative urgency, global impulsivity, and experiential avoidance. We ran various structural models to maximize model fit, which were informed by modification indices including freed factor loadings and inclusion of additional covariances. No models fully met conventional standards for best fit.1 Results that follow are from the best fitting model with the greatest parsimony (ꭓ^2^ = 402.16, df = 14, p < 0.001; RMSEA = 0.22; SRMR = 0.11; CFI = 0.86; TLI = 0.71; See Table 5 and Fig. 2 for graphical summaries of results). There was a significant direct effect of trait anxiety on experiential avoidance (β = 0.62). Trait anxiety was a significant predictor of both negative urgency (β = 0.49) and the latent factor of global impulsivity (β = 0.43). Negative urgency was not a significant predictor of experiential avoidance (β = −0.23); however, the global impulsivity factor did significantly predict experiential avoidance (β = 0.57). The indirect effect of trait anxiety on experiential avoidance through negative urgency was not significant (β = −0.11). The indirect effect through global impulsivity was small (β = 0.24) and approaching significance (p = 0.06). Both the total indirect effect (β = 0.13) and the total effect (β = 0.74) were significant. These findings do not support our hypothesis that when anxiety is high, negative urgency would be more strongly related to experiential avoidance than global impulsivity.Table 5Results from primary structural equation model.Table 5STD95% CINegative UrgencySTAIT → NU0.49[0.42, 0.56] *NU → EA-0.23[− 0.77, 0.04]Indirect Effect-0.11[− 0.37, 0.02]Impulsivity Latent FactorSTAIT → IMP0.43[0.31, 0.54] *IMP → EA0.57[0.30, 1.11] *Indirect Effect0.24[0.11, 0.52] *Direct & Total EffectsDirect Effect0.62[0.52, 0.70] *Total Indirect Effect0.13[0.06, 0.21] *Total Effect0.75[0.71, 0.78] *Note: STAIT = Trait Anxiety; NU = Negative Urgency; EA = Experiential Avoidance; IMP = Impulsivity Latent Factor; 95% CI = 95% confidence intervals for standardized coefficients; STD = standardized coefficients; * = p < 0.05. 95% confidence intervals derived via bias corrected bootstrap procedure with 10,000 random draws.Fig. 2Graphical model of primary structural equation model.Fig. 2
Due to poor fit of our primary SEM with our data, we sought to further understand relationships between variables of interest using a secondary SEM including data-driven latent factors. Item-level data from both the UPPS-P and BIS-11 were subjected to an EFA to agnostically examine the factor structure for indicators of impulsivity. A six-factor solution was specified for an EFA with maximum likelihood extraction with an orthogonal varimax rotation. This model demonstrated good fit (ꭓ^2^ = 7707.01, df = 3397, p < 0.001; RMSEA = 0.05; SRMR = 0.03; See Supplementary Table 1 for full results of item-level EFA). Factor 1 was comprised of 11 items from the UPPS-P (lack of) premeditation subscale (i.e., Premeditation; factor 0.35 - 0.65). Factor 2 was comprised of 25 items and reflected the UPPS-P negative and positive urgency subscales (i.e., Urgency; factor 0.42 - 0.81). Factor 3 was comprised of 12 items from the UPPS-P sensation seeking subscale (i.e. Sensation Seeking; factor 0.36 – 0.78). Factor 4 was comprised of 8 items from the UPPS-P perseverance subscale (i.e., Perseverance; factor 0.39 - 0.62).
Factor 5 was comprised of 13 items from the BIS-11 related to future planning (i.e., Planning; factor 0.35 – 0.67). Factor 5 reflected similar themes to Factor 1 (premeditation) and they were strongly correlated (r = 0.74). However, only four items cross-loaded > 0.30 (range: 0.31–0.34) between Factor 1 and Factor 5. Factor 6 was comprised of 16 items and reflected themes of hyperactivity (e.g., restless, fidgeting, motor activity) parallel to the “Hyperactivity” factor in a three factor structure for ADHD [57] (i.e., Hyperactivity; factor 0.32–0.67). Four items of the BIS-11 did not significantly load onto any factors and were excluded from subsequent analyses. These data-driven factors of impulsivity were included as intermediary variables in a subsequent SEM.
We conducted a secondary SEM with the 6 factors of impulsivity identified through the above EFA as intermediary variables (instead of negative urgency and global impulsivity as done in the primary SEM). In keeping with the primary model, paths were specified from trait anxiety (i.e., independent variable) to experiential avoidance (i.e., dependent variable), between trait anxiety and the intermediary impulsivity factors, and between these impulsivity factors and experiential avoidance.
In an effort to optimize fit of the secondary SEM, multiple models were tested informed by modification indices. Notably, in the best fitting secondary model, we also specified a bifactor model for experiential avoidance informed by an item-level EFA on the AAQ-II. This model did not meet conventional standards of good fit (ꭓ^2^ = 12431.66, df = 4161, p < 0.001; RMSEA = 0.06; CFI = 0.77, TLI = 0.76; SRMR = 0.18), suggesting that additional constructs may be needed to explain variance in this model (Fig. 3; Table 6). As in our primary SEM, there was a direct effect of trait anxiety on experiential avoidance (β = 0.51). Trait anxiety was a significant predictor of the urgency factor (β = 0.24), the sensation seeking factor (β = −0.10), the perseverance factor (β = 0.38), the planning factor (β = 0.44), and the hyperactivity factor (β = 0.27). Significant predictors of experiential avoidance included the urgency factor (β = 0.24), the planning factor (β = −0.15), and the hyperactivity factor (β = 0.23). There were significant indirect effects of three out of our six impulsivity factors; urgency (β = 0.06) and hyperactivity (β = 0.06) demonstrated small positive indirect effects, such that individuals who experienced high anxiety and high urgency or hyperactivity demonstrated greater experiential avoidance. In contrast, the (lack of) planning factor demonstrated a small negative indirect effect (β = −0.06), such that individuals who exhibited high anxiety and less planning tendencies demonstrated less experiential avoidance. Both the total indirect effect (β = 0.07) and the total effect (β = 0.59) of the overall model were statistically significant as well.Fig. 3Graphical model of secondary structural equation model.Fig. 3Table 6Results from secondary structural equation model.Table 695% CISTDFactor Premeditation (F1)STAIT → F1[− 0.03, 0.14]0.06F1 → EA[− 0.20, 0.10]-0.05Indirect Effect[− 0.01, 0.01]-0.003Factor Urgency (F2)STAIT → F2[0.16, 0.31]*0.27F2 → EA[0.10, 0.38]*0.28Indirect Effect[0.02, 0.10]0.08Factor Sensation Seeking (F3)STAIT → F3[− 0.19, − 0.02]-0.11F3 → EA[− 0.13, 0.08]-0.02Indirect Effect[− 0.01, 0.01]0.002Factor Perseverance (F4)STAIT → F4[0.31, 0.45]0.41F4 → EA[− 0.08, 0.18]0.06Indirect Effect[− 0.03, 0.07]0.02Factor Planning (F5)STAIT → F5[0.37, 0.51]0.47F5 → EA[− 0.27, − 0.01]-0.2Indirect Effect[− 0.12, − 0.004]-0.1Factor Hyperactivity (F6)STAIT → F6[0.19, 0.34]*0.3F6 → EA[0.09, 0.38]*0.27Indirect Effect[0.42, 0.60]0.08Direct & Total EffectsDirect Effect[0.42, 0.60]*0.65Total Indirect Effect[0.00, 0.15]0.08Total Effect[0.53, 0.64]*0.73Note: STAIT = Trait Anxiety; NU = Negative Urgency; EA = Experiential Avoidance; 95% CI = 95% confidence intervals for standardized coefficients; STD = standardized coefficients; * = p < 0.05. 95% confidence intervals derived via bias corrected bootstrap procedure with 10,000 random draws.
The present study contributes to an ever-growing body of transdiagnostic literature by examining complex relationships between dimensions of impulsivity, anxiety and experiential avoidance. As an emotion-based dimension of impulsivity, negative urgency was of particular interest in these analyses given its previous links to both anxiety [58], [59], [60], [61], [62] and experiential avoidance [35], [38]. Consistent with our hypothesis, there were positive relationships between anxiety and experiential avoidance, negative urgency, and global impulsivity. Contrary to our hypothesis, negative urgency was not sufficient to explain the role of impulsive traits in anxiety-based experiential avoidance.
Previous research examining the factor structures of measures of impulsivity like the UPPS-P have proposed revised factors reflecting nuances in the assessment of urgency [63]. Our exploratory secondary model, revealed six factors of impulsivity based on items from the UPPS-P and BIS-11. Three of these factors (i.e., premeditation, sensation seeking, and perseverance) were consistent with subscales of the UPPS-P [46]. In addition, an emotion-based impulsivity factor included items from both urgency subscales. Items from the BIS-11 were largely captured by the Planning and Hyperactivity Factors, consistent with previous work that has proposed a bifactor structure for this measure [64]. The items on the Planning Factor reflected lack of control, while the items on the Hyperactivity Factor reflected reflexive action.
Our secondary CFA including these data-driven factors provided nuance to our initial finding that negative urgency was not sufficient to explain the role of impulsive traits in experiential avoidance. We found that individuals with greater anxiety and emotion-based impulsivity – regardless of emotional valence – demonstrated greater experiential avoidance. Our data-driven model also revealed that individuals who exhibited greater anxiety and hyperactivity also demonstrated greater experiential avoidance. By contrast, high anxiety and less planning (which reflects greater impulsivity) were associated with less experiential avoidance.
Anxiety has previously been linked to emotion based impulsivity, regardless of emotional valence [65], [66]. Our findings also cohere with the Reflexive Responding to Emotion Framework (RRE) which positions emotion-based impulsivity as a combination of heightened emotional responsivity and behavioral disinhibition [21], [67]. Within this framework, behavioral disinhibition in response to emotion may manifest as either impulsive action or inaction [68]. Greater negative urgency has previously been associated with greater use of emotion regulation strategies rooted in inaction (e.g., disengagement; [69]). Therefore, this link between broad emotion-based impulsivity and experiential avoidance may reflect a reflexive lack of behavior (i.e. inaction) in response to any heightened emotion.
We expected that non-emotional dimensions of impulsivity would show weaker associations with experiential avoidance than emotional dimensions. However, we found that two non-emotional dimensions of impulsivity, hyperactivity and lack of planning, had opposing relationships with experiential avoidance when anxiety was high. It is possible that motor-based impulsivity (e.g., restlessness, fidgeting) may serve as a way to cope with internal distress or may be an outward manifestation of it. Furthermore, in keeping with the RRE framework, it is possible that this finding may reflect a relationship between heightened emotion and reflexive action. Overall, our results corroborate previous findings linking anxiety to both emotional and motor-based dimensions of impulsivity [70], [71].
Our finding that greater planning was linked to greater experiential avoidance in more anxious participants was surprising given that greater emotion regulation, reflection, and mindfulness is linked to reduced experiential avoidance [72]. A speculative interpretation could be that individuals who are anxious and lack planning tendencies may be less reflective, or less likely to link their anxiety to the types of internal cues (e.g., emotion, physical sensations) that are avoided by people with high experiential avoidance. Notably, the premeditation factor, which reflected similar themes to the planning factor, demonstrated a similar positive relationship with anxiety, negative relationship with experiential avoidance, and a negative indirect effect. To our knowledge, there is no prior literature linking experiential avoidance to lack of premeditation or planning. Findings from this research underscore the complexity of the relationships between anxiety and impulsivity and provide avenues for future work using larger samples and additional measures.
Interpretation of findings from these analyses is limited primarily by suboptimal model fit of confirmatory models. Poor model fit likely reflects a combination of factors including cross-loadings between variables, issues of validity with self-report measures included in these analyses, variance not captured by these measures, low power and floor effects of item responses. Although the BIS-11 has been widely used to measure impulsivity [44], a prior meta-analysis found that items did not map on to the three subscales (i.e., attention, motor, and non-planning) proposed [45]. Furthermore, prior work has suggested that the trait anxiety subscale of the STAI [49] and the AAQ-II [50] may better capture overall negative affect than solely anxiety or experiential avoidance respectively. These issues of measure validity suggest that findings may reflect relationships between impulsivity and broader negative affect as opposed to anxious avoidance. Although good model fit is particularly important for interpretation of causal models, prior research has suggested that fit indices are less conceptually relevant to the types of questions tested in the current study, examining cross sectional relationships between observed and latent variable [73], [74]. We attempted to account for the limitations of these measures by employing multiple indicators of impulsivity and conducting agnostic, item level factor analyses. However, future research may use the well-validated Multidimensional Experiential Avoidance Questionnaire (MEAQ; [75]), or Behavioral Inhibition System/Behavioral Approach System Scale (BIS/BAS; [76], [77]). Additionally, poor model fit suggests that factors not captured by these measures likely contribute to unexplained variance. We would expect observed relationships within this model to be stronger in future research if factors such as anxiety sensitivity [24], [78], and externalizing psychopathology (e.g., substance use, ADHD) are accounted for as covariates. Finally, data collection for these analyses was not restricted to clinical samples. Although subscale scores across all of our measures spanned clinical and subclinical ranges, individual items - particularly on measures of impulsivity - demonstrated low endorsement rates. Findings from this unrestricted sample would likely benefit from replication in a sample with a greater range of symptom endorsement.
Future studies may also reexamine these relationships within a longitudinal cohort to better understand causal links between anxious and impulsive constructs and in samples of individuals who hold sexual, racial, or gender minority identities. Individuals from these backgrounds may exhibit elevated rates of psychopathology [79], and may engage in risky behaviors commonly associated with impulsivity [80]. The unique psychosocial stressors (e.g., racism, homophobia) and lack of structural support that individuals with minoritized identities face may exacerbate relationships between negative urgency and experiential avoidance.
The present study demonstrated that the relationship between anxiety and experiential avoidance is differentially related to various dimensions of impulsivity. Although negative urgency alone did not account for significant variance in the relationship between anxiety and experiential avoidance, emotion-based impulsivity and hyperactivity did. Conversely, we found that greater anxiety and low planning tendencies were associated with less experiential avoidance. Results from this research provide a foundation to examine individual differences in profiles of co-occurring anxiety and impulsivity that could yield different targets for behavioral treatments.
Participant payment was funded by the Rackham Graduate Student Research Grant through the 10.13039/100006801Rackham Graduate School and the Bordin Award through the Clinical Science Area in the Department of Psychology at the 10.13039/100007270University of Michigan. Training in statistical methods used in this study was funded by the Rackham Summer Training Award through the Institute for Social Research at the University of Michigan.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.