A Taxometric Analysis of Experiential Avoidance

BETH-00995; No of Pages 13; 4C: Available online at www.sciencedirect.com ScienceDirect

Behavior Therapy xx (xxxx) xxx

www.elsevier.com/locate/bt

A Taxometric Analysis of Experiential Avoidance

Alex Kirk University of Colorado Boulder Joshua J. Broman-Fulks Appalachian State University Joanna J. Arch University of Colorado Boulder University of Colorado Cancer Center

EXPERIENTIAL AVOIDANCE (EA) IS defined as an Experiential avoidance, a trait-like construct referring to the unwillingness to confront and remain in contact tendency to rigidly avoid or change unpleasant internal with unpleasant thoughts and emotions, coupled experiences stemming from an unwillingness to experience with attempts to change the form, frequency, or them, is believed to contribute to the development and occurrence of those experiences and the situations maintenance of various forms of psychopathology. Despite that elicit them (Hayes, Wilson, Gifford, Follette, & significant research on this construct, it remains unclear Strosahl, 1996). EA has largely been conceptualized whether experiential avoidance is dimensional or categorical and measured as trait-like (Gámez, Chmielewski, at the latent level. The current study examined the latent Kotov, Ruggero, & Watson, 2011), encompassing structure of experiential avoidance using three taxometric a variety of beliefs and behaviors that promote analytic approaches (MAXimum Eigenvalue, Mean Above avoidance strategies in the context of distressing Minus Below A Curve, Latent-Mode Factor Analysis) situations. Thus, individuals high in EA are more applied to data from two independent samples and using likely to engage in avoidance behaviors (e.g., three widely used measures of experiential avoidance. The leaving a room containing a feared stimulus, first sample (n = 922) completed the Multidimensional distracting oneself from sadness) when distress or Experiential Avoidance Questionnaire (Gámez, Chmie- a distressing situation arises or is anticipated. lewski, Kotov, Ruggero, & Watson, 2011), while the second Importantly, while avoidance itself is not intrinsi- sample (n = 615) completed the Brief Experiential Avoid- cally pathological, it can become problematic when ance Questionnaire (Gámez et al., 2014) and Acceptance it is both contextually insensitive and relied upon in and Action Questionnaire-II (Bond et al., 2011). Across a chronic manner as a means of regulating both samples and all three measures, experiential avoidance unpleasant internal experiences (Forsyth, Eifert, & exhibited a dimensional structure. The clinical and research Barrios, 2006), such as in the case of consistently implications of this finding for experiential avoidance are high EA. Indeed, this is reflected in the consistent discussed. finding that higher EA is associated with negative outcomes including greater mental health symptom severity (Gámez et al., 2011; Kashdan et al., 2014; Keywords: experiential avoidance; taxometrics; MEAQ; BEAQ; Thompson & Waltz, 2010), lower quality of life AAQ-II (Gámez et al., 2011; Kashdan, Morina, & Priebe, 2009; Kirk, Meyer, Whisman, Deacon, & Arch, 2019), and poorer physical health (Andrew & Address correspondence to Alex Kirk, M.A., University of Dulin, 2007; Berghoff, Tull, DiLillo, Messman- Colorado Boulder, 345 Muenzinger, Boulder, CO 80309-0345; Moore, & Gratz, 2017). Further, high EA has been e-mail: [email protected]. shown to be associated with a variety of diagnoses

0005-7894/© 2020 Association for Behavioral and Cognitive Therapies. including anxiety disorders (Bardeen, Fergus, & Published by Elsevier Ltd. All rights reserved. Orcutt, 2013; Kashdan et al., 2014; Newman &

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 2 kirk et al.

Llera, 2011; Thompson & Waltz, 2010), depres- individuals can express varying levels of EA, such as sion (Spinhoven, Drost, de Rooij, van Hemert, & “high” or “low” EA (e.g., Kashdan et al., 2014). Penninx, 2014), alcohol and substance use disor- Importantly, when used in this way, “high” or ders (Levin et al., 2012; Shorey et al., 2017), and “low” EA does not reflect qualitatively distinct borderline personality disorder (Jacob, Ower, & types of EA, but rather a shared type of EA with Buchholz, 2013). Taken together, EA can be different levels of severity. Thus, the set of conceptualized as a latent psychological factor experiences and behaviors between “high” and that influences a range of avoidance behaviors in “low” EA individuals is similar, but the frequency distressing contexts and across diverse psycholog- or intensity of those experiences or behaviors differ. ical symptom profiles. However, despite this common description of EA, An important finding is that EA does not appear there is insufficient psychometric evidence to to be unidimensional. Though early measures of indicate that EA operates along a single continuum EA assessed it as a single factor, including the of behavior. Indeed, it may be the case that EA Acceptance and Action Questionnaire-II (AAQ-II; functions categorically such that different scores on Bond et al., 2011), later factor analytic studies EA measures reflect qualitatively distinct categories. on measures of EA have found evidence for a Viewed this way, those who are high versus low EA multidimensional structure (Gámez et al., 2011). would then reflect qualitatively distinct groups Specifically, the Multidimensional Experiential characterized by divergent sets of behavior, rather Avoidance Questionnaire (MEAQ; Gámez et al., than a shared set of behaviors that differ only in the 2011) identified six factors of EA, including frequency or intensity with which they are behavioral avoidance, distress aversion, procrasti- expressed. An example of this might be that low nation, distraction/suppression, repression/denial, EA could reflect an adaptive coping style, while and distress endurance. As such, EA appears to be high EA might reflect more pathological avoidance comprised of distinct forms of trait-like avoidance observed in chronic and severe mental health tendencies and does not necessarily operate in the disorders. Such categorical distinctions among same way among different people. Thus, two varying levels of avoidance behaviors have been individuals expressing high levels of EA could discussed within the literature on safety behaviors express distinct dimensions of EA, which influence versus adaptive coping (e.g., Arnaudova, Kindt, avoidance behaviors differently. Importantly, the Fanselow, & Beckers, 2017; Thwaites & Freeston, MEAQ has withstood psychometric scrutiny, 2005), though no clear consensus exists. Elucidat- showing evidence that the overall scale and ing the underlying dimensional or categorical subscales are robust and diverge from constructs structure of EA advances both our theoretical such as distress or negative affectivity (Rochefort, understanding of this construct and our approach Baldwin, & Chmielewski, 2018). The MEAQ also to measuring it. comes in a shortened version, the Brief Experiential Central to determining the categorical or dimen- Avoidance Questionnaire (BEAQ; Gámez et al., sional structure of a construct is the use of taxometric 2014), which retains the same dimensions mea- analyses, a set of statistical procedures that allows sured by the MEAQ though only a single overall researchers to mathematically evaluate the latent score is derived. The AAQ (Hayes et al., 2004)and structure of a given construct (Beauchaine, 2007; AAQ-II (Bond et al., 2011) reflected groundbreak- Meehl, 1995). Through use of taxometrics, con- ing theoretical developments on EA (Hayes et al., structs can be shown to have taxonic (i.e., dichoto- 1996; Hayes et al., 2004) and remain widely used, mous, categorical) or nontaxonic (i.e., continuous, though have not held up as well to psychometric dimensional) structures (Meehl & Golden, 1982). scrutiny (Rochefort et al., 2018; Wolgast, 2014). Taxometric analyses have been applied to various Despite accumulating data on the importance of psychological diagnoses including anxiety disorders EA in predicting critical outcomes, less is known (Kollman, Brown, Liverant, & Hofmann, 2006), about its latent structure. Early work on EA has posttraumatic stress disorder (Broman-Fulks et al., proposed that it exists within a dysfunctional range 2006), obsessive-compulsive disorder (Olatunji, Wil- of normal behavior, implying a more dimensional liams, Haslam, Abramowitz, & Tolin, 2008), major conceptualization (Hayes et al., 1996). That is, depression (Prisciandaro & Roberts, 2005; individuals may exhibit varying levels of EA, but Solomon, Ruscio, Seeley, & Lewinsohn, 2006), only at a sufficiently high level does it influence bipolar disorder (Ahmed, Green, Clark, Stahl, & negative outcomes. Consistent with this dimension- McFarland, 2011; Prisciandaro & Roberts, 2011), al view, researchers have historically referred to EA and eating disorders (Olatunji et al., 2012). Beyond in terms of a given self-report score that exists along formal diagnoses, various studies have also exam- a broader range of scores. For example, different ined the latent structure of mental health constructs

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 taxometrics of experiential avoidance 3 that, like EA, predict the onset and exacerbation of Despite accumulating research delineating the mental health disorders. For example, past research function and importance of EA in various clinical has found evidence of dimensional structures for outcomes, no known studies have investigated the intolerance of uncertainty (Carleton et al., 2012), latent structure of EA to determine whether it distress tolerance (Stevens, Kertz, Bjorgvinsson, & operates as a dimensional or categorical construct. McHugh, 2018), worry (Olatunji, Broman-Fulks, To address this gap, the present research examined Bergman, Green, & Zlomke, 2010), and fear of the latent structure of EA through use of taxometrics negative evaluation (Weeks, Norton, & Heimberg, to determine whether it operates dimensionally or 2009). While the majority of psychological diag- categorically. Two studies were conducted by noses and mental health constructs tend to operate applying three nonredundant taxometric procedures dimensionally (Haslam, Holland, & Kuppens, to data collected from two independent samples 2012), it remains important to identify categorical using three separate measures of EA (i.e., MEAQ, constructs in order to better understand the BEAQ,AAQ-II).Study1soughttoestablishthe development and etiology of those constructs, latent structure of EA using indicators representing and to better inform assessment and treatment the subcomponents of EA measured by the MEAQ (Beauchaine, 2007). Indeed, studies assessing the (i.e., Behavioral Avoidance, Distress Aversion, Pro- latent structure of important etiological constructs crastination, Distraction/Suppression, Repression/ are sometimes mixed. For example, taxometric Denial, and Distress Endurance), allowing for a analyses of anxiety sensitivity have yielded both more nuanced examination of the structure of EA dimensional (Broman-Fulks et al., 2010)and across multiple dimensions. Study 2 applied taxo- categorical (Bernstein et al., 2007) results. Clarify- metric analyses to indicators derived from the BEAQ ing such inconsistencies or uncertainty regarding and AAQ-II to determine whether findings from the latent structure of a given construct is impor- Study 1 would extend to the other commonly used tant for both the measurement of that construct as measures of EA. As EA has been treated theoretically well as our conceptualization regarding etiology as a dimensional construct, operating along a and treatment. continuum of normal human behavior, it was An accurate understanding of a construct’slatent hypothesized that EA would evidence a dimensional structure is important for several reasons (Meehl, structure as measured by the six components of the 1992). For example, the aim of assessment instru- MEAQ, as well as single sum scores from the BEAQ ment development and administration varies de- and AAQ-II. pending on a variable’s latent structure (Beauchaine, 2007; Ruscio & Ruscio, 2002; Walters & Ruscio, Study 1: Taxometric Analysis of the MEAQ 2009). Whereas a categorical model of EA would methods imply that assessment instruments should be de- Participants signed to sort individuals into their valid categorical Participants consisted of 1,094 individuals who class, a dimensional model suggests that measures completed the survey through Prolific, a crowd- should aim to assess and discriminate across the full working platform similar to Amazon’s Mechanical spectrum of EA (Grove, 1991). Creating artificial Turk (MTurk) (Palan & Schitter, 2018). Previous dichotomizations along a dimensional construct research using these platforms has shown greater would lead to an unnecessary loss of information diversity when compared to university samples, and and reduction in statistical power. Similarly, know- comparable reliability (Buhrmester, Kwang, & ing the latent structure of EA would inform the Gosling, 2011; Chandler & Shapiro, 2016). Partic- appropriate use of language to communicate about ipants were paid $2.85 for completing the study the subject. For example, if EA is categorical, it measures, which were part of a larger assessment would be appropriate to refer to individuals as battery that took an average of 33 minutes to “having” or “not having” EA, whereas a dimen- complete. sional approach would suggest that reference to To be eligible for the study, participants had to be levels or ranges along the continuum of EA would be at least 18 years of age, reside in the United States, more relevant. Knowing the latent structure of EA report English as their native language, and also has the potential to inform research into the complete the full survey in a valid manner. To etiology and maintenance of EA. Whereas a cate- complete the survey in a valid manner, participants gorical latent structure suggests a discrete etiology could not provide irrelevant or off-topic answers, (e.g., a specific genetic source, environmental cause, and had to respond correctly to validity items and particular gene-environment interaction), a dimen- finish the survey in a reasonable amount of time sional structure would indicate a multiply deter- (i.e., between 8 minutes and 5 hours) based on pilot mined (e.g., additive, graded) etiology. testing that established the minimum amount of

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 4 kirk et al. time needed to respond to survey content. Based on nal consistency in the current sample at α =0.93 these eligibility criteria, 172 were excluded due to and α = 0.91, respectively. not completing the full survey (n = 110), completing Data Analytic Strategy the survey in an unrealistically fast amount of time Taxometric analyses were conducted in a manner implying a lack of content responsiveness (n = 40), consistent with the recommendations of Ruscio, providing irrelevant or off-topic responses (n = 16), Walters, Marcus, and Kaczetow (2010) and by using failing to pass attention checks inquiring their their R programs (Ruscio & Wang, 2017). First, location (n = 3), not reporting English as their potential indicators of an EA taxon were identified native language (n = 2), or not residing in the United and subjected to preliminary analyses to determine States (n = 1). Upon excluding these participants, a whether they met minimum recommended criteria final sample of 922 participants was used for and were suitable for taxometric analysis. Specifical- subsequent analyses. Final sample participants ly, six indicators were selected, representing the six were predominantly male (52.7%), White/Cau- MEAQ subscales, in an effort to provide compre- casian (77.4%), and reported an average age hensive representation of the conceptualized EA of 36.27 (SD = 11.57). The ethnic diversity of construct. Given the absence of a suggested base the sample included 7.5% identifying as multiple rate of pathological EA in the literature to guide case races or ethnicities, 5.5% as Black/African assignment for taxometric analyses, prevalence rates American, 5.4% as Asian American, 2.6% as of emotion-related psychopathology known to be Hispanic or Latino/a, 0.4% as Native American, associated with elevated EA were used to inform and 0.1% as Native Hawaiian/Pacific Islander, putative taxon and complement case allocation (e.g., with 0.1% preferring not to provide ethnicity Kessler et al., 2005). Specifically, indicator scores information. were summed, and cases scoring in the top 25% were Measures assigned to the conjectured taxon group for prelim- Multidimensional Experiential Avoidance Ques- inary (CheckData) and taxometric analyses (Run- tionnaire (MEAQ). The MEAQ (Gámez et al., 2011) Taxometrics). The sample and indicators were is a 62-item, multidimensional measure of EA. Each selected for taxometric analysis to meet recommend- item is rated on a Likert-type scale ranging from 1 ed standards for taxometric analysis (Meehl, 1995, (strongly agree)to6(strongly disagree). Based on 1999; Waller & Meehl, 1998), which include: a its original validation and factor analysis, the sample size of N N 300, a putative taxon group base MEAQ assesses EA across six domains: Behavioral rate of at least N =50andp ≥ .10, a minimum of at Avoidance (e.g., “If I am in a slightly uncomfort- least two potential indicators, each indicator is able situation, I try to leave right away”), Distress comprised of at least four ordered categories (C ≥ Aversion (e.g., “IwishIcouldgetridofallofmy 4), each indicator demonstrates between-group negative emotions”), Procrastination (e.g., “Itend validity of at least d ≥ 1.25, and relatively low to put off unpleasant things that need to get done”), within-group correlations among indicators (rwg ≤ Distraction/Suppression (e.g., “I usually try to .30). Although these are generally considered ideal distract myself when I feel something painful”), conditions for taxometric analyses, it should be Repression/Denial (e.g., “Idon’trealizeI’manx- noted that simulation studies have indicated that the ious until other people tell me”), and Distress taxometric procedures are robust and can produce Endurance (e.g., “Even when I feel uncomfortable, accurate results even when using data that fall Idon’t give up working toward things I value”). A somewhat outside of the recommended ranges or recent psychometric study provided strong and fail to meet one or more of these criteria, particularly independent support for the MEAQ as a valid when the characteristics of some of the variables are measure of EA (Rochefort et al., 2018), showing more favorable (Ruscio, Ruscio, & Carney, 2011). promising construct validity in areas where other Thus, indicators in the present study that met measures of EA have loaded more onto distress recommended cutoffs for these criteria, as deter- factors (Wolgast, 2014). Within the current sam- mined by the CheckData function, were retained and ple, each subscale from the MEAQ showed good to cases reassigned using summed scores across the excellent internal consistency (αs = 0.89 – 0.93). retained indicators. Next, the retained indicators were To assess the anxiety and depression levels of the submitted to the RunTaxometrics function, which sample, two widely used measures, the Generalized applies three independent taxometric procedures, Anxiety Disorder-7 (GAD-7; Spitzer, Kroenke, MAXimum Eigenvalue (MAXEIG; Waller & Meehl, Williams, & Löwe, 2006)andPatient Health 1998), Mean Above Minus Below A Curve (MAM- Questionnaire-8 (PHQ-8; Kroenke et al., 2009), BAC; Meehl & Yonce, 1994), and Latent-Mode were administered, and both exhibited high inter- Factor Analysis (L-Mode; Waller & Meehl, 1998), to

Table 1 Table 2 Descriptive Statistics for the MEAQ Pearson Correlation Coefficients for the Full Sample in Study 1 Measures Mean (SD) Range Scale 1 2 3 4 5 6 MEAQ, Behavioral Avoidance 38.74 (11.68) 11 – 66 1. Behavioral – 0.72 0.60 0.52 0.49 -0.44 MEAQ, Distress Aversion 46.39 (13.94) 13 – 78 Avoidance MEAQ, Procrastination 24.81 (8.10) 7 – 42 2. Distress 0.72 – 0.45 0.58 0.40 -0.30 MEAQ, Distraction/Suppression 28.25 (7.25) 7 – 42 Aversion MEAQ, Repression/Denial 34.17 (12.24) 13 – 73 3. Procrastination 0.60 0.45 – 0.27 0.45 -0.45 MEAQ, Distress Endurance 47.63 (9.48) 11 – 66 4. Distraction/ 0.52 0.58 0.27 – 0.24 -0.01 MEAQ, Total Score 201.73 (45.51) 74 – 362 Suppression – Note. Means, standard deviations, and ranges for Multidimen- 5. Repression/ 0.49 0.40 0.45 0.24 -0.17 sional Experiential Avoidance Questionnaire (MEAQ) subscales Denial and total score. 6. Distress -0.44 -0.30 -0.45 -0.01 -0.17 – Endurance the research data. Each procedure was run using the Note. All results statistically significant at p b .001 except for the correlation between Distress Endurance and Distraction/Suppres- standard settings in RunTaxometrics, which uses sion (p N .05). 50 windows in MAMBAC and MAXEIG with 90% overlap, and indicators are assigned to be used in all possible input-output pairings (MAM- with community adult scores (Gámez et al., 2011). BAC) or triplets (MAXEIG), rather than creating Intercorrelations among all MEAQ subscales are summed input variables. Cases were assigned to shown in Table 2. In terms of distress, the sample putative taxon and complement classes using a endorsed moderate symptoms of generalized anx- base rate of 0.35, which was selected based on the iety based on GAD-7 scores (M = 13.57, SD = 5.64) relation between EA and psychopathology (e.g., (Spitzer et al., 2006), and moderately severe Chawla & Ostafin, 2007; Hayes, Strosahl, & symptoms of depression based on PHQ-8 scores Wilson, 1999) and the relatively high rates of (M = 15.35, SD = 6.12) (Kroenke et al., 2009). anxiety and depressive symptomology endorsed by the current sample (see below). In addition, parallel MEAQ Taxometric Analyses analyses of comparison categorical and dimension- Preliminary analyses were conducted using the al data are performed. The resulting output allows CheckData function to examine whether the six judgments to be made regarding the classification indicators representing the six subscales of the of plots using the Comparison Curve Fit Index MEAQ met the recommended suitability criteria (CCFI), which compares the averaged curve outlined above. Results indicated that three of the generated by each of the three taxometric proce- MEAQ indicators, representing the Distress Endur- dures with simulated taxonic and dimensional ance, Procrastination, and Distraction/Suppression curves generated using data that match the general subscales, failed to meet the minimum validity characteristics of the research data, including skew, criteria (i.e., indicator validity b 1.25 SD) and were and kurtosis (Ruscio, Ruscio, & Meron, 2007). dropped from subsequent analyses. The remaining CCFI scores range between zero and one, with three indicators, representing the Behavioral Avoid- scores greater than .55 being indicative of categor- ance, Distress Aversion, and Repression/Denial ical structure, scores less than .45 being indicative subscales of the MEAQ, met all other suitability of dimensional structure, and scores between .45 criteria, including demonstrating relatively low and .55 being considered ambiguous. nuisance correlations (taxon M = .22, complement M = .29), and were therefore retained for the results taxometric analysis. Descriptive statistics for the MEAQ are provided in When submitted to MAMBAC, MAXEIG, and Table 1. In terms of descriptive results from the L-Mode analyses, the three indicators generated a MEAQ, no clear benchmarks exist allowing for mean CCFI score of .31 (MAMBAC = .17, classifications of EA severity based on MEAQ MAXEIG = .36, L-Mode = .41). As can be seen in scores. However, based on its original validation, Figure 1, the averaged curves generated by the the current sample endorsed scores representing the research data were more consistent with those midpoint between community adults and psychiat- generated by simulated dimensional data than the ric patients with the exception of the Distraction/ simulated taxonic data. Thus, the results provide Suppression subscale, which was more consistent preliminary support indicating that EA, as mea- with psychiatric patient scores, and the Distress sured by the MEAQ, is a dimensionally distributed Endurance subscale, which was more consistent phenomenon.

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 6 kirk et al.

MAMBAC MAMBAC Categorical Comparison Data Dimensional Comparison Data Mean Difference Mean Difference 0.7 0.9 1.1 1.3 0.7 0.9 1.1 1.3

0 1020304050 0 1020304050

Cut Cut

MAXEIG MAXEIG Categorical Comparison Data Dimensional Comparison Data Eigenvalue Eigenvalue 0.15 0.25 0.35 0.15 0.25 0.35

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Score Score

LMode LMode Categorical Comparison Data Dimensional Comparison Data Density Density 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4

-3 -2 -1 0 1 2 3 -3 -2 -1 0 1 2 3

Factor Score Factor Score

FIGURE 1 Taxometric Analyses of the MEAQ Note. Average MAMBAC (top), MAXEIG (middle), and L-Mode (bottom) curves for the three indicator solution imposed on simulated taxonic (left) and dimensional (right) comparison curves.

Study 2: Taxometric Analysis of the BEAQ and information and completed the BEAQ, AAQ-II, and AAQ-II GAD-7 (Note: Due to aiming for a briefer survey, the PHQ-8 was not administered in this study). In an effort to extend the findings of Study 1 using Participants were paid $0.25 for completing all different measures of EA, data were also collected measures, which took approximately 5 minutes to from a second, independent MTurk sample using complete. To participate, MTurk participants had two other commonly used measures of experiential to be at least 18 years of age, reside in the United avoidance, the BEAQ and AAQ-II. States, report English as their native language, methods respond accurately to validity check items, and Participants complete the full survey. After being informed Participants consisted of 615 individuals who were of inclusion criteria, participants completed a recruited from MTurk and provided demographic demographic questionnaire, and any prospective

MAMBAC MAMBAC Categorical Comparison Data Dimensional Comparison Data Mean Difference Mean Difference 0.60.81.01.2 0.60.81.01.2

01020304050 0 1020304050

Cut Cut

MAXEIG MAXEIG Categorical Comparison Data Dimensional Comparison Data Eigenvalue Eigenvalue 0.15 0.25 0.35 0.15 0.25 0.35

-1.0 -0.5 0.0 0.5 1.0 1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Score Score

LMode LMode Categorical Comparison Data Dimensional Comparison Data Density Density 0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4

-2 -1 0 1 2 3 -2 -1 0 1 2 3

Factor Score Factor Score

FIGURE 2 Taxometric Analyses of the BEAQ Note. Average MAMBAC (top), MAXEIG (middle), and L-Mode (bottom) curves for the Seven Brief Experiential Avoidance Questionnaire (BEAQ) indicators imposed on simulated taxonic (left) and dimensional (right) comparison curves. participant that indicated they did not meet any of Latino/a 3.4%, Asian 2.9%, multiple ethnicities the participation criteria (e.g., reported age b 18) 2.9%, Native American 0.3% and Native Hawai- were immediately informed that they were ineligible ian/Pacific Islander 0.2%, with 0.8% preferring not to participate in the study and thanked for their to provide ethnicity information. interest. Of the 691 prospective participants that began the survey, a total of 615 provided complete Measures data and were included in all analyses. Participants Brief Experiential Avoidance Questionnaire were predominantly female (61.8%), White/Cau- (BEAQ). The BEAQ is a shortened version of casian (83.6%), and had an average age of 38.91 the MEAQ, consisting of 15 items. The BEAQ items (SD = 13.32). Other ethnicities represented includ- reflect the same six dimensions of EA as the MEAQ. ed: Black/African American 5.9%, Hispanic or In contrast to the MEAQ, however, the BEAQ was

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 8 kirk et al. developed to be used solely as a single sum score items from the Behavioral Avoidance subscale, two across all subscales rather than using individual from the Distress Aversion subscale, and one from subscales on their own (Gámez et al., 2014). The the Distraction/Suppression subscale. Results indi- BEAQ showed good internal consistency within the cated that the three taxometric procedures generated current sample (α = 0.86). a mean CCFI score of .32 (MAMBAC = .24, MAXEIG = .33, L-Mode = .39), and, as can be Acceptance and Action Questionnaire-II (AAQ-II). seen in Figure 2, the data plots were consistent with The AAQ-II is a unidimensional, 7-item measure of those produced by simulated dimensional data with psychological inflexibility. However, although not comparable data characteristics. Thus, these findings explicitly focusing on EA as the sole construct of were consistent with those of Study 1 and provide interest, the creators of the AAQ-II designed it to be further support suggesting that EA is a dimensional used as a measure of EA (Bond et al., 2011), and it is construct. commonly used as one. Each item is rated on a AAQ-II Taxometric Analyses Likert-type scale ranging from 1 (never true)to7 Given the purported unifactorial nature of the AAQ- (always true). Importantly, more recent analyses of II, the 7 individual items comprising the AAQ-II were the AAQ-II indicate that it may in fact be a stronger selected as prospective indicators. Preliminary anal- measure of negative emotionality than of EA yses revealed that all seven items met minimum (Rochefort et al., 2018; Wolgast, 2014). However, validity criteria (M = 2.23). Nuisance correlations the AAQ-II remains widely used as a measure of between the indicators were somewhat higher than EA, and thus was included in the current study to the generally recommended cutoffs (taxon M =.39, evaluate its latent structure. The AAQ-II showed complement M = .46). However, given the favor- strong internal consistency within the current ability of the other data parameters and research sample (α =0.95). suggesting that taxometric analyses tend to be robust Similar to Study 1, the GAD-7 was administered under such circumstances (e.g., Ruscio et al., 2011), to assess anxiety levels in the current sample. the seven indicators were submitted to the three Again, the GAD-7 demonstrated strong internal taxometric procedures. Results indicated that the consistency (α =0.93). research data generated a mean CCFI score of .31 results (MAMBAC = .27, MAXEIG = .35, L-Mode = .31). The correlation between the BEAQ (M =44.73,SD= As depicted in Figure 3, the plots produced by the 12.89) and AAQ-II (M = 20.76, SD = 10.26) was AAQ-II indicators were consistent with those gener- medium to large, r =.59,p b .001. Similar to the first ated by simulated dimensional data with comparable sample, the sample from Study 2 reported moderate data characteristics. These findings indicate that the symptoms of generalized anxiety disorder based on construct(s) measured by the AAQ-II appear to be scores from the GAD-7 (M =13.11,SD =5.89) dimensional, and if one accepts the AAQ-II as a (Spitzer et al., 2006). measure of EA, then these results provide further evidence that EA is continuous, rather than categor- BEAQ Taxometric Analyses ical, at the latent level. As research has suggested that the BEAQ is not suitable for creating subscales comparable to those General Discussion available on the MEAQ (Gámez et al., 2014), the Over the past couple decades, considerable research 15 individual items on the BEAQ served as has accumulated indicating that EA serves as a prospective indicators for the present study. This vulnerability factor for a variety of mental health was done in an effort to retain indicators that concerns, ranging from mood and anxiety disorders represent the full spectrum of EA in the absence of to substance abuse (e.g., Cribb, Moulds, & Carter, subscale scores. 2006; Gratz, Bornovalova, Delany-Brumsey, Nick, Preliminary analyses of the suitability of the 15 & Lejuez, 2007; Kashdan, Barrios, Forsyth, & BEAQ indicators revealed that eight of the indicators Steger, 2006). Although it appears that researchers did not meet minimum validity criteria (d b 1.25) and have generally operated on the assumption that EA were removed from further analysis. The remaining is dimensional, rather than categorical, at the latent seven indicators (BEAQ item numbers: 2, 8, 11, 12, level, the present investigation represents the first 13, 14, 15) demonstrated low nuisance correlations attempt to empirically inform which approach is (taxon M =.21,complementM = .25), met all other most appropriate given the nature of EA. Across suitability criteria, and were submitted to taxometric three separate measures of EA, three taxometric analysis using the same parameters described in analytic approaches, and two independent samples, Study 1. Of these seven indicators, four represented results converged on EA functioning as a dimensional

MAMBAC MAMBAC Categorical Comparison Data Dimensional Comparison Data Mean Difference Mean Difference 0.8 1.2 1.6 0.8 1.2 1.6

0 1020304050 0 1020304050

Cut Cut

MAXEIG MAXEIG Categorical Comparison Data Dimensional Comparison Data Eigenvalue Eigenvalue 0.1 0.2 0.3 0.4 0.1 0.2 0.3 0.4

-1.0 -0.5 0.0 0.5 1.0 1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Score Score

LMode LMode Categorical Comparison Data Dimensional Comparison Data Density Density 0.0 0.2 0.4 0.0 0.2 0.4

-2 -1 0 1 2 3 -2-10123

Factor Score Factor Score

FIGURE 3 Taxometric Analyses of the AAQ-II Note. Average MAMBAC (top), MAXEIG (middle), and L-Mode (bottom) curves for the seven Acceptance and Action Questionnaire-II (AAQ-II) indicators imposed on simulated taxonic (left) and dimensional (right) comparison curves. latent factor. These findings provide initial empirical represent descriptions of ranges along a single support for the dimensional conceptualization of EA. dimension, with differences reflecting variations in Although this is the first study to examine the intensity or frequency of avoidance tendencies latent structure of EA through use of taxometric rather than types. Importantly, viewing EA as a analyses, these results align with existing theoretical dimensional construct has implications relevant to understanding of EA (Hayes et al., 1996). That is, research and clinical work. First, a dimensional EA appears to function in a manner such that view suggests caution in trying to draw meaningful, individuals may vary in their level of EA, from low categorical classifications of EA. For example, to high, but such descriptors do not denote separate attempts to dichotomize EA into nonclinical and categories of EA with clear demarcations between clinical subgroups would be contraindicated and them. Rather, terms like “high EA” and “low EA” should be avoided. Instead, instruments assessing

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 10 kirk et al.

EA should strive to reliably determine where a given offering further support for EA functioning as a individual stands within the broad spectrum of EA dimensional behavioral construct. However, it is magnitude or intensity. Thus, rather than incorpo- important to restate that the AAQ-II indicators rating items on a measure of EA that aim to exhibited higher cutoffs for nuisance correlations distinguish separate categorical groups, items than is generally recommended. Because of this, should instead aim to discriminate lower or higher results from the AAQ-II warrant more caution in levels of EA across its spectrum (Walters & Ruscio, their interpretation as compared to the MEAQ and 2009). Statistical analysis of EA should similarly BEAQ. That stated, given evidence suggesting avoid dichotomization and instead make use of the limited overlap in what the AAQ and MEAQ/ full range of EA datapoints to avoid losing BEAQ actually measure (Rochefort et al., 2018), information or distorting the underlying construct. the AAQ-II results do not detract from the clear In addition, these results support using factor findings of the MEAQ/BEAQ analyses suggesting analytic data to determine which factors are central that EA is dimensional. to a given construct as to better inform taxometric These results further align with the majority of analyses, particularly when that construct has been psychological constructs, which have been found to viewed as unidimensional and multidimensional in operate dimensionally upon taxometric analyses. structure in separate measures. Second, as previous Indeed, EA is conceptualized as a trait-like construct work has pointed out, relative to categorical that can contribute to psychological disorder at constructs, dimensional constructs tend to have sufficiently high levels. Other trait-like vulnerability many etiological factors (Haslam, 1997). This factors, such as intolerance of uncertainty (Carleton suggests that EA is likely to develop due to a et al., 2012) and disgust sensitivity (Olatunji & complex, prolonged interaction of multiple envi- Broman-Fulks, 2007), both of which influence the ronmental and/or genetic factors. Thus, future development of various anxiety-related disorders, research investigating the etiology of EA should have similarly been found to operate dimensionally. work to identify the different variables that These results place EA in a similar category as these collectively contribute to the development of EA other constructs, operating dimensionally and con- at the individual level. Finally, the results from the tributing to negative mental health outcomes at higher current study support more recent calls for greater levels, though cautioning against the use of arbitrary attention on clinically relevant, dimensional con- cut-offs to introduce qualitative classifications. structs that are largely transdiagnostic in nature, as The current study had several notable strengths. EA has been traditionally viewed (Hayes et al., Consistent with the recommendations of Meehl 1996). Importantly, these constructs offer the (1995) and Ruscio et al. (2010), the present research potential to better inform treatment approaches employed a rigorous multiple hurdles consistency that seek to focus on processes of therapeutic testing approach. Indicators derived from the three change, and to address the underlying pathological most commonly used measures of EA and across two mechanisms that create, maintain, or exacerbate independent samples were submitted to three math- mental health concerns (e.g., Hayes & Hofmann, ematically distinct taxometric procedures. Only 2017). Thus, future work can seek to understand indicators that exceeded recommended validity cut- the various ways in which EA manifests and offs were used, and the indicators demonstrated low functions across different diagnostic presentations nuisance correlations. The fact that the collective in order to better target this shared etiological results of these multiple analyses converged on a mechanism. By doing so, interventions may become dimensional solution without exception enhances more effective in both the prevention and treatment confidence in the findings. In addition, consistent of all psychological disorders that are in part driven with modern recommendations for taxometric re- by higher levels of EA. search, the present study relied on an objective fit Based on the results from Study 2, results from measure for interpreting the data plots that resulted both the BEAQ and AAQ-II converged with those from the taxometric analyses rather than subjective of the MEAQ from Study 1. Because both the inspection (Ruscio et al., 2010). Collectively, these BEAQ and AAQ-II were used as single sum score strengths enhance confidence in claims that the latent measures per validation recommendations, individ- structure of EA is nontaxonic or dimensional at the ual items were used as indicators. Based on the latent level. BEAQ, and similar to Study 1, six of the seven Although this research has several important indicators were derived from the Behavioral Avoid- strengths, conclusions should be drawn with consid- ance and Distress Aversion subscale. Importantly, eration of the study limitations. First, because these the results from the BEAQ and AAQ-II from Study samples came from online platforms, the clinical 2 were consistent with the results from Study 1, status of participants could not be confirmed as they

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 taxometrics of experiential avoidance 11 would be in face-to-face settings. This limitation Andrew, D. H., & Dulin, P. L. (2007). The relationship between could be overcome by advancing this work in diverse, self-reported health and mental health problems among older adults in New Zealand: Experiential avoidance as a face-to-face samples. Second, taxometric analyses moderator. Aging & Mental Health, 11(5), 596–603. are only effective when appropriate indicators are https://doi.org/10.1080/13607860601086587 submitted to analysis. Although the present research Arnaudova, I., Kindt, M., Fanselow, M., & Beckers, T. (2017). relied on indicators derived from the most commonly Pathways towards the proliferation of avoidance in anxiety and used measures of EA, there appears to be concern implications for treatment. Behaviour Research and Therapy, 96,3–13. https://doi.org/10.1016/j.brat.2017.04.004 expressed by many in the literature regarding the Bardeen, J. R., Fergus, T. A., & Orcutt, H. K. (2013). most appropriate assessment tools for measuring EA. Experiential avoidance as a moderator of the relationship For example, although EA has been most commonly between anxiety sensitivity and perceived stress. Behavior assessed using the AAQ-II, and theoretical frame- Therapy, 44(3), 459–469. https://doi.org/10.1016/j.beth. works (e.g., ACT) have been considerably influenced 2013.04.001 Beauchaine, T. P. (2007). A brief taxometrics primer. Journal of by the results of this research, recent psychometric Clinical Child and Adolescent Psychology, 36(4), 654–676. studies have indicated that the AAQ-II is, at best, a https://doi.org/10.1080/15374410701662840 suboptimal measure of EA, and at worst, predom- Berghoff, C. R., Tull, M. T., DiLillo, D., Messman-Moore, T., inantly a measure of distress and neuroticism & Gratz, K. L. (2017). The role of experiential avoidance in (Rochefort et al., 2018; Wolgast, 2014). The the relation between anxiety disorder diagnoses and future physical health symptoms in a community sample of young MEAQ and BEAQ have been developed to address adult women. Journal of Contextual Behavioral Science, 6 these concerns, though the longitudinal predictive (1), 29–34. https://doi.org/10.1016/j.jcbs.2016.11.002 utility of these measures for psychopathological Bernstein, A., Zvolensky, M. J., Norton, P. J., Schmidt, N. B., outcomes has not been thoroughly explored. Given Taylor, S., Forsyth, J. P., … Cox, B. (2007). Taxometric and these considerations, indicators derived from current factor analytic models of anxiety sensitivity: integrating approaches to latent structural research. Psychological measures of EA may be suboptimal, and it is possible Assessment, 19(1), 74–87. https://doi.org/10.1037/1040- that future research employing other indicators of EA 3590.19.1.74 (e.g., biometric measures, direct observation of Bond, F. W., Hayes, S. C., Baer, R. A., Carpenter, K. M., avoidance behavior) could generate different out- Guenole, N., Orcutt, H. K., … Zettle, R. D. (2011). comes. Replication of the present findings using Preliminary psychometric properties of the Acceptance and Action Questionnaire-II: a revised measure of psychological novel measures of EA or a multimethod approach to inflexibility and experiential avoidance. Behavior Therapy, assessing EA would help to enhance confidence in the 42(4), 676–688. https://doi.org/10.1016/j.beth.2011.03.007 dimensional conclusion. Broman-Fulks, J., Deacon, B., Olatunji, B. O., Bondy, C. L., conclusion Abramowitz, J. S., & Tolin, D. F. (2010). Categorical or dimensional: A reanalysis of the anxiety sensitivity con- This is the first known study to examine the latent struct. Behavior Therapy, 41(2), 154–171. https://doi.org/ structure of EA through taxometric analyses. Using 10.1016/j.beth.2009.02.005 three taxometric procedures applied to two large Broman-Fulks,J.J.,Ruggiero,K.J.,Green,B.A.,Kilpatrick, D. G., Danielson, C. K., Resnick, H. S., & Saunders, B. E. samples, the current study provided convergent (2006). Taxometric Investigation of PTSD: Data from two evidence that EA operates dimensionally rather nationally representative samples. Behavior Therapy, 37 than categorically. Future work should extend this (4), 364–380. https://doi.org/10.1016/j.beth.2006.02.006 approach to diverse samples to elucidate the precise Buhrmester, M., Kwang, T., & Gosling, S. D. (2011). Amazon's relationship between EA and negative mental Mechanical Turk: A new source of inexpensive, yet high- quality, data? Perspectives on Psychological Science, 6(1), health outcomes. Given the strong support for a 3–5. https://doi.org/10.1177/1745691610393980 dimensional conceptualization of EA, it will be Carleton, R. N., Weeks, J. W., Howell, A. N., Asmundson, G. J. important that researchers constructing theoretical G., Antony, M. M., & McCabe, R. E. (2012). Assessing the models, assessment measures, and interventions latent structure of the intolerance of uncertainty construct: An targeting EA align their language, approach, and initial taxometric analysis. Journal of Anxiety Disorders, 26 (1), 150–157. https://doi.org/10.1016/j.janxdis.2011.10.006 methodology to be consistent with the observed Chandler, J., & Shapiro, D. (2016). Conducting clinical dimensional latent structure of EA. research using crowdsourced convenience samples. Annual Conflict of Interest Statement Review of Clinical Psychology, 12,53–81. https://doi.org/ 10.1146/annurev-clinpsy-021815-093623 The authors declare that there are no conflicts of interest. Chawla, N., & Ostafin, B. (2007). Experiental avoidance as a functional dimensional approach to psychopathology: An References empirical review. Journal of Clinical Psychology, 63, 871–890. https://doi.org/10.1002/jclp.20400 Ahmed, A. O., Green, B. A., Clark, C. B., Stahl, K. C., & Cribb, G., Moulds, M. L., & Carter, S. (2006). Rumination and McFarland, M. E. (2011). Latent structure of unipolar and experiential avoidance in depression. Behaviour Change, 23 bipolar mood symptoms. Bipolar Disorders, 13(5-6), (3), 165–176. https://doi.org/10.1375/bech.23.3.165 522–536. https://doi.org/10.1111/j.1399-5618.2011. Forsyth, J. P., Eifert, G. H., & Barrios, V. (2006). Fear 00940.x conditioning in an emotional regulation context: A fresh

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008 12 kirk et al.

perspective on the origins of anxiety disorders. In M.G. interaction and daily interactions of people with social Craske, D. Hermans, & D. Vansteenwegen (Eds.), Fear and anxiety disorder. Emotion, 14(4), 769–781. https://doi. learning: From basic processes to clinical implications org/10.1037/a0035935 (pp. 133–153). Washington, DC: American Psychological Kashdan, T. B., Morina, N., & Priebe, S. (2009). Post-traumatic Association. stress disorder, social anxiety disorder, and depression in Gámez, W., Chmielewski, M., Kotov, R., Ruggero, C., Suzuki, survivors of the Kosovo War: Experiential avoidance as a N., & D., W. (2014). The Brief Experiential Avoidance contributor to distress and quality of life. Journal of Anxiety Questionnaire: Development and initial validation. Psycho- Disorders, 23(2), 185–196. https://doi.org/10.1016/j. logical Assessment, 26(1), 35–45. https://doi.org/10.1037/ janxdis.2008.06.006 a0034473 Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, Gámez, W., Chmielewski, M., Kotov, R., Ruggero, C., & K. R., & Walters, E. E. (2005). Lifetime Prevalence and Watson, D. (2011). Development of a measure of experien- Age-of-Onset Distributions of DSM-IV Disorders in the tial avoidance: The Multidimensional Experiential Avoid- National Comorbidity Survey Replication. Archives of ance Questionnaire. Psychological Assessment, 23(3), General Psychiatry, 62(6), 593. https://doi.org/10.1001/ 692–713. https://doi.org/10.1037/a0023242.supp archpsyc.62.6.593 Gratz, K. L., Bornovalova, M. A., Delany-Brumsey, A., Nick, Kirk, A., Meyer, J. M., Whisman, M. A., Deacon, B. J., & Arch, B., & Lejuez, C. W. (2007). A laboratory-based study of the J. J. (2019). Safety behaviors, experiential avoidance, and relationship between childhood abuse and experiential anxiety: A path analysis approach. Journal of Anxiety avoidance among inner-city substance users: The role of Disorders, 64,9–15. https://doi.org/10.1016/j.janxdis.2019. emotional nonacceptance. Behavior Therapy, 38(3), 03.002 256–268. https://doi.org/10.1016/j.beth.2006.08.006 Kollman, D. M., Brown, T. A., Liverant, G. I., & Hofmann, Grove, W. M. (1991). When is a diagnosis worth making? A S. G. (2006). A taxometric investigation of the latent statistical comparison of two prediction strategies. Psycho- structure of social anxiety disorder in outpatients with logical Reports, 68,3–17. https://doi.org/10.1001/archinte. anxiety and mood disorders. Depression and Anxiety, 23 166.10.1092 (4), 190–199. https://doi.org/10.1002/da.20158 Haslam, N. (1997). Evidence that male sexual orientation is a Kroenke, K., Strine, T. W., Spitzer, R. L., Williams, J. B., Berry, matter of degree. Journal of Personality and Social J. T., & Mokdad, A. H. (2009). The PHQ-8 as a measure of Psychology, 73(4), 862–870. https://doi.org/10.1037/0022- current depression in the general population. Journal of 3514.73.4.862 Affective Disorders, 114(1-3), 163–173. https://doi.org/10. Haslam, N., Holland, E., & Kuppens, P. (2012). Categories 1016/j.jad.2008.06.026 versus dimensions in personality and psychopathology: A Levin, M. E., Lillis, J., Seeley, J., Hayes, S. C., Pistorello, J., & quantitative review of taxometric research. Psychological Biglan, A. (2012). Exploring the relationship between Medicine, 42(5), 903–920. https://doi.org/10.1017/ experiential avoidance, alcohol use disorders, and alcohol- S0033291711001966 related problems among first-year college students. Journal Hayes, S. C., & Hofmann, S. G. (2017). The third wave of of American College Health, 60(6), 443–448. https://doi. cognitive behavioral therapy and the rise of process-based org/10.1080/07448481.2012.673522 care. World Psychiatry, 16(3), 245–246. https://doi.org/10. Meehl, P. E. (1992). Factors and taxa, traits and types, 1002/wps.20442 differences of degree and differences in kind. Journal of Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (1999). Personality, 60(1), 117–174. https://doi.org/10.1111/j. Acceptance and commitment therapy: An experiential 1467-6494.1992.tb00269.x approach to behavior change. New York: Guilford Press. Meehl, P. E. (1995). Bootstraps taxometrics. Solving the Hayes, S. C., Strosahl, K., Wilson, K. G., Bissett, R. T., classification problem in psychopathology. American Psy- Pistorello, J., Dosheen, T., … McCurry, S. M. (2004). chologist, 50(4), 266–275. https://doi.org/10.1037//0003- Measuring experiential avoidance: A preliminary test of a 066x.50.4.266 working model. The Psychological Record, 54, 553–578. Meehl, P. E. (1999). Clarifications about taxometric method. https://doi.org/10.1007/BF03395492 Applied and Preventive Psychology, 8(3), 165–174. https:// Hayes, S. C., Wilson, K. G., Gifford, E. V., Follette, V. M., & doi.org/10.1016/S0962-1849(05)80075-7 Strosahl, K. (1996). Experiential avoidance and behavioral Meehl, P. E., & Golden, R. R. (1982). Taxometric methods. In disorders: A functional dimensional approach to diagnosis J. N. Butcher & P. C. Kendall (Eds.), The handbook of and treatment. Journal of Consulting and Clinical Psychol- research methods in clinical psychology (pp. 127–181). New ogy, 64(6), 1152–1168. https://doi.org/10.1037//0022- York: Wiley. 006x.64.6.1152 Meehl, P. E., & Yonce, L. J. (1994). Taxometric analysis: I. Jacob, G. A., Ower, N., & Buchholz, A. (2013). The role of Detecting taxonicity with two quantitative indicators using experiential avoidance, psychopathology, and borderline means above and below a sliding cut (MAMBAC proce- personality features in experiencing positive emotions: A dure). Psychological Reports, 74(3, Pt 2), 1059–1274. path analysis. Journal of Behavior Therapy and Experimen- Newman, M. G., & Llera, S. J. (2011). A novel theory of tal Psychiatry, 44(1), 61–68. https://doi.org/10.1016/j.jbtep. experiential avoidance in generalized anxiety disorder: 2012.07.006 A review and synthesis of research supporting a contrast Kashdan, T. B., Barrios, V., Forsyth, J. P., & Steger, M. F. avoidance model of worry. Clinical Psychology Review, 31 (2006). Experiential avoidance as a generalized psycholog- (3), 371–382. https://doi.org/10.1016/j.cpr.2011.01.008 ical vulnerability: Comparisons with coping and emotion Olatunji, B. O., & Broman-Fulks, J. J. (2007). A taxometric study regulation strategies. Behaviour Research and Therapy, 44 of the latent structure of disgust sensitivity: Converging (9), 1301–1320. https://doi.org/10.1016/j.brat.2005.10.003 evidence for dimensionality. Psychological Assessment, 19 Kashdan, T. B., Goodman, F. R., Machell, K. A., Kleiman, (4), 437–448. https://doi.org/10.1037/1040-3590.19.4.437 E.M.,Monfort,S.S.,Ciarrochi,J.,&Nezlek,J.B. Olatunji, B. O., Broman-Fulks, J. J., Bergman, S. M., Green, B.A., (2014). A contextual approach to experiential avoidance & Zlomke, K. R. (2010). A taxometric investigation of the and social anxiety: Evidence from an experimental latent structure of worry: Dimensionality and associations

with depression, anxiety, and stress. Behavior Therapy, 41(2), ance, distress tolerance, and substance use cravings among 212–228. https://doi.org/10.1016/j.beth.2009.03.001 adults in residential treatment for substance use disorders. Olatunji, B. O., Broman-Fulks, J. J., Ciesielski, B. G., Journal of Addictive Diseases, 36(3), 151–157. https://doi. Zawilinski, L. L., Shewmaker, S., & Wall, D. (2012). A org/10.1080/10550887.2017.1302661 taxometric investigation of the latent structure of eating Solomon, A. R. I., Ruscio, J., Seeley, J. R., & Lewinsohn, P. M. disorders. Psychiatry Research, 197(1-2), 97–102. https:// (2006). A taxometric investigation of unipolar depression in doi.org/10.1016/j.psychres.2011.12.016 a large community sample. Psychological Medicine, 36(7), Olatunji, B. O., Williams, B. J., Haslam, N., Abramowitz, J. S., 973–985. https://doi.org/10.1017/S0033291706007689 & Tolin, D. F. (2008). The latent structure of obsessive- Spinhoven,P.,Drost,J.,deRooij,M.,vanHemert,A.M.,& compulsive symptoms: A taxometric study. Depression and Penninx, B. W. (2014). A longitudinal study of experi- Anxiety, 25(11), 956–968. https://doi.org/10.1002/da. ential avoidance in emotional disorders. Behavior Ther- 20387 apy, 45(6), 840–850. https://doi.org/10.1016/j.beth.2014. Palan, S., & Schitter, C. (2018). Prolific.ac—Asubjectpool 07.001 for online experiments. Journal of Behavioral and Experi- Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Löwe, B. mental Finance, 17,22–27. https://doi.org/10.1016/j.jbef. (2006). A brief measure for assessing generalized anxiety 2017.12.004 disorder: The GAD-7. JAMA Internal Medicine, 166(10), Prisciandaro, J. J., & Roberts, J. E. (2005). A taxometric 1092–1097. https://doi.org/10.1001/archinte.166.10.1092 investigation of unipolar depression in the national comor- Stevens, K. T., Kertz, S. J., Bjorgvinsson, T., & McHugh, R. K. bidity survey. Journal of Abnormal Psychology, 114(4), (2018). Investigating the latent structure of distress intoler- 718–728. https://doi.org/10.1037/0021-843x.114.4.718 ance. Psychiatry Research, 262, 513–519. https://doi.org/ Prisciandaro, J. J., & Roberts, J. E. (2011). Evidence for the 10.1016/j.psychres.2017.09.036 continuous latent structure of mania in the Epidemiologic Thompson, B. L., & Waltz, J. (2010). Mindfulness and Catchment Area from multiple latent structure and construct experiential avoidance as predictors of posttraumatic stress validation methodologies. Psychological Medicine, 41(3), disorder avoidance symptom severity. Journal of Anxiety 575–588. https://doi.org/10.1017/S0033291710001078 Disorders, 24(4), 409–415. https://doi.org/10.1016/j. Rochefort, C., Baldwin, A. S., & Chmielewski, M. (2018). janxdis.2010.02.005 Experiential avoidance: An examination of the construct Thwaites, R., & Freeston, M. H. (2005). Safety-seeking validity of the AAQ-II and MEAQ. Behavior Therapy, 49 behaviours: Fact or function? How can we clinically (3). https://doi.org/10.1016/j.beth.2017.08.008 differentiate between safety behaviours and adaptive coping Ruscio, J., & Ruscio, A. M. (2002). A structure-based approach strategies across anxiety disorders? Behavioural and Cogni- to psychological assessment: Matching measurement models tive Psychotherapy, 33(2), 177–188. https://doi.org/10. to latent structure. Assessment, 9,4–16. https://doi.org/10. 1017/s1352465804001985 1177/1073191102091002 Waller, N. G., & Meehl, P. E. (1998). Multivariate taxometric Ruscio, J., Ruscio, A. M., & Carney, L. M. (2011). Performing procedures: Distinguishing types from continua. Thousand taxometric analysis to distinguish categorical and dimen- Oaks, CA: Sage Publications. sional variables. Journal of Experimental Psychopathology, Walters, G. D., & Ruscio, J. (2009). To sum or not to sum: 2(2), 170–196. https://doi.org/10.5127/jep.010910 Taxometric analysis with ordered categorical assessment Ruscio, J., Ruscio, A.M., & Meron, M. (2007). Applying the items. Psychological Assessment, 21,99–111. https://doi. bootstrap to taxometric analysis: Generating empirical org/10.1037/a0015010 sampling distributions to help interpret results. Multivariate Weeks, J. W., Norton, P. J., & Heimberg, R. G. (2009). Behavioral Research, 42(2), 349–386. https://doi.org/10. Exploring the latent structure of two cognitive components 1080/00273170701360795 of social anxiety: taxometric analyses of fears of negative Ruscio, J., Walters, G. D., Marcus, D. K., & Kaczetow, W. and positive evaluation. Depression and Anxiety, 26(2), (2010). Comparing the relative fit of categorical and E40–E48. https://doi.org/10.1002/da.20414 dimensional latent variable models using consistency tests. Wolgast, M. (2014). What does the Acceptance and Action Psychological Assessment, 22(1), 5–21. https://doi.org/10. Questionnaire (AAQ-II) really measure? Behavior Therapy, 1037/a0018259 45(6), 831–839. https://doi.org/10.1016/j.beth.2014.07.002 Ruscio, J., & Wang, S. B. (2017). RTaxometrics: Taxometric analysis (Version 2.3). Retrieved from https://CRAN.R- project.org/package=RTaxometrics RECEIVED: November 14, 2019 Shorey, R. C., Gawrysiak, M. J., Elmquist, J., Brem, M., ACCEPTED: April 10, 2020 Anderson, S., & Stuart, G. L. (2017). Experiential avoid- AVAILABLE ONLINE: xxxx

Please cite this article as: A. Kirk, J. J. Broman-Fulks and J. J. Arch, A Taxometric Analysis of Experiential Avoidance, Behavior Therapy, https://doi.org/10.1016/j.beth.2020.04.008