Fundamental Fears in Greece (I)
The Psychometric Properties of the Greek Adaptation of the Anxiety Sensitivity Index-3
Abstract
Abstract:Introduction: The present study examined the psychometric properties of a Greek adaptation of the Anxiety Sensitivity Index-3 (ASI-3; Taylor et al., 2007). Method: We translated the ASI-3 following a forward-backward method and then, in addition to measures of anxiety and depression (DASS-21; Lovibond & Lovibond, 1995; Lyrakos et al., 2011), we administered it to a nonclinical general population sample (N = 611) recruited online. Results: Confirmatory factor analysis revealed that a bifactor model with three orthogonal group factors best fit the data, followed by a correlated three-factor model. An examination of the dimensionality of the ASI-3 and the reliability of its dimensions suggested the presence of a reliable, strong AS general factor and comparatively weaker group factors. The ASI-3 appears to measure AS invariantly across gender. We report preliminary evidence for its convergent, discriminant, and divergent validity. Conclusion: The Greek adaption of the ASI-3 revealed adequate psychometric properties. Future studies should explore its criterion-related validity by administering the Greek adaptation of the ASI-3 to clinical samples and explore its relationship to other key constructs of anxiety sensitivity’s nomological network.
Introduction
Although present in the psychological literature for more than a century (e.g., Freud, 1895), anxiety sensitivity (AS, i.e., the fear of anxiety) was most prominently emphasized in the expectancy model of fear, panic, and anxiety (Reiss, 1991; Reiss & McNally, 1985). AS reflects an individual’s predisposition to catastrophically appraise anxiety based on beliefs that its symptoms have harmful consequences (Reiss & McNally, 1985). Consequently, AS is viewed as a trait-like construct that acts as a central diathesis for fearfulness; it has been likened to Eysenck’s concept of dysthymia (Rachman, 1991). Contemporary evidence regarding the relevance of AS in anticipation of threat corroborates this view (Nelson et al., 2015). Results from empirical studies have consistently highlighted the importance of AS in fear- and anxiety-related pathology. Specifically, AS has been associated with the exacerbation and/or development of panic disorder (Kim et al., 2017; Maller & Reiss, 1992; Poletti et al., 2015; Rassovsky et al., 2000; Schmidt & Cook, 1999; Zvolensky et al., 2001), generalized anxiety disorder and worry (e.g., Floyd et al., 2005; Knapp et al., 2016; Ruiz, 2014), social anxiety (Allan et al., 2018; Laposa et al., 2015; Panayiotou et al., 2014), posttraumatic stress disorder (Bernstein & Zvolensky, 2007; Elwood et al., 2009; Feldner et al., 2007), and obsessive-compulsive disorder (e.g., Blakey et al., 2017; Wheaton et al., 2012). AS has also been implicated in conditions outside the strict anxiety domain, including substance abuse (e.g., Castellanos-Ryan et al., 2013) and depression (Allan et al., 2014, 2018; Rosellini et al., 2011).
Yet, AS is distinct from high-order traits such as trait anxiety (McNally, 1996; Reiss, 1997) and the experience of negative affect (McNally, 2002): The source of distress encapsulated in the AS construct is one’s own anxiety and not a diverse range of potential stressors. This distinction hints at a qualitative difference between higher-order traits that reflect an individual’s accrued negative affectivity over time, which are thus not explanatory (Claridge & Davis, 2001; Ormel et al., 2004), and lower-order traits, such as AS, which facilitate fearfulness and consequently result in negative affect (Carleton, 2016).
The heritability of AS has been researched and generally supported (Jang et al., 1999; Stein et al., 1999; Zavos et al., 2012). Environmental influences on the development of AS include modeling, vicarious conditions, and stressful life events (McLaughlin & Hatzenbuehler, 2009; Schmidt et al., 2000; Watt et al., 1998). Preoccupied and fearful attachment styles also mark elevated AS (Weems et al., 2002). There are mixed findings regarding gender differences in AS, with some studies showing higher AS scores in women than in men (Armstrong & Khawaja, 2002; Deacon et al., 2002; Ghisi et al., 2016; Rifkin et al., 2015; Sandin et al., 2007; Stewart et al., 1997). This finding aligns with the wider anxiety literature (Lewinsohn et al., 1998; McLean et al., 2011). Others, however, report nonsignificant differences between men and women (Escocard et al., 2008; Osman et al., 2010; Taylor et al., 2007).
Measuring AS
The first widely used measure of AS was the Anxiety Sensitivity Index (ASI), first introduced by Reiss et al. (1986) and based on the Epstein-Reiss Fear of Anxiety Scale (FAS; Epstein, 1982). Although the construct was initially conceptualized as unidimensional (Reiss & McNally, 1985), the researchers made no hypothesis about the dimensionality of the scales, and after they had administered it to two samples of college students, principal components analysis revealed a predominantly unidimensional measure. However, later factor analytic studies of the ASI showed inconsistent results, with some supporting a unidimensional solution (Peterson & Heilbronner, 1987; Reiss et al., 1988; Sandin et al., 1996; Taylor et al., 1992), while others supported multidimensional solutions (Schmidt & Joiner, 2002; Telch et al., 1989; Wardle et al., 1990; Zinbarg et al., 1997). This disagreement might have been because of different factor retention criteria, small sample sizes, and the original unidimensional conceptualization of AS (Taylor et al., 2007). To remedy this, Taylor and Cox (1998a, 1998b) created the 36-item Anxiety Sensitivity Index–Revised (ASI-R) and the 60-item Anxiety Sensitivity Profile (ASP), both of which served to measure six distinct AS domains, each with an equal number of items. However, both the ASI-R and the ASP continued yielding inconsistent factor structures across studies (Armstrong et al., 2006; Arnau et al., 2009; Deacon et al., 2003; Schmidt et al., 2008; Van der Does et al., 2003; Zvolensky et al., 2003).
Subsequently, Taylor et al. (2007) developed the Anxiety Sensitivity Index-3 (ASI-3), based on the ASI-R item pool, to measure the three most replicated dimensions of AS: physical concerns (e.g., “It scares me when my heart beats rapidly.”), cognitive concerns (e.g., “When I cannot keep my mind on a task, I worry that I might be going crazy.”), and social concerns (e.g., “It is important for me not to appear nervous.”). In the ASI-3, one measures each dimension by an equal number of items conceptually related only to their respective dimension. The ASI-3’s psychometric properties were examined on a large multinational dataset comprising both clinical and nonclinical samples from earlier studies using the ASI-R. The results revealed that a three-factor model demonstrated a better fit than two-factor and one-factor alternatives. The three-factor solution was then replicated in six subsamples. Furthermore, the ASI-3 measured AS invariantly across gender. Subsequent studies have replicated the three-factor solution in clinically and ethnically diverse samples (Cai et al., 2018; Farris et al., 2015; Hilton et al., 2022; Kemper et al., 2012; Lim & Kim, 2012; Sandin et al., 2007; Wheaton et al., 2012).
Given the stable three-factor structure, researchers used the ASI-3 to examine the correlates of domain-specific AS, consistently and incrementally associating the social concerns domain with social anxiety (Allan et al., 2014, 2018; Olthuis et al., 2014; Taylor et al., 2007; Wheaton et al., 2012), while associating the physical concerns domain with panic, SAD, specific phobia, and health anxiety (Allan et al., 2014; Fergus & Bardeen, 2013; Fetzner et al., 2014; Olthuis et al., 2014; Taylor et al., 2007; Wheaton et al., 2012). Lastly, researchers have associated the cognitive concerns subscale with depression and symptoms of GAD, OCD, and PTSD (Allan et al., 2014; 2018; Olthuis et al., 2014; Taylor et al., 2007).
Bifactor Models of the ASI-3
Bifactor models of the ASI-3 have also been examined, consistently yielding a better fit than the originally proposed hierarchical/correlated-factors solution (Allan et al., 2015; Ebesutani et al., 2014; Ghisi et al., 2016; Jardin et al., 2018; Osman et al., 2010; Rifkin et al., 2015). Measurement invariance of the ASI-3 across gender also found support using a bifactor model (Ebesutani et al., 2014; Ghisi et al., 2016).
Bifactor models are particularly relevant when addressing issues of dimensionality and reliability (Reise et al., 2010; Rodriguez et al., 2016a, 2016b). If a general factor explains a dominant proportion of the variance in item responses, despite the existence of multiple group factors, a measure can be considered unidimensional and can lend itself to be modeled as such with trivially biased general factor loadings (Reise et al., 2007). Furthermore, a measure may be multidimensional, but this does not mean that group factors are reliable above and beyond the general factor, nor that items adequately specify the group factors they reflect in a structural equation modeling (SEM) context. Accordingly, Rodriguez et al. (2016b) conducted a large investigation of data from studies employing bifactor models to examine the psychometric properties of 50 multidimensional measures (including two studies on the ASI-3; Ebesutani et al., 2014; Osman et al., 2010). They concluded that (a) unit-weighted composite scores reflecting a general factor are generally reliable and robust to bias because of multidimensionality, (b) unit-weighted composite scores reflecting group factors rarely explain sufficient variance above and beyond the general factor, (c) group factors often lack determinacy and construct reliability (Gorsuch, 1980; Grice, 2001; Hancock & Mueller, 2001), and (d) many multidimensional measures can be modeled as unidimensional with relatively little bias. Results specific to the ASI-3, corroborated by later investigations (Ghisi et al., 2016; Jardin et al., 2018) lent themselves to similar conclusions.
The Present Study
Because of the substantive importance of AS across the psychopathology spectrum and Ghisi et al.’s (2016) call for more examinations of the ASI-3 across cultures, this study proposed to evaluate the psychometric properties of the ASI-3 in a Greek general population sample, using a newly adapted version. Given the many updated versions of the original ASI created and tested throughout the years, we decided to adapt the ASI-3 to the Greek language because of its stable factor structure and widespread use. Regarding the factor structure of the Greek version of the ASI-3 (from here on referred to simply as the ASI-3), we hypothesized that (H1) a bifactor model with three orthogonal group factors would best fit the data, followed by a three-correlated-factors model (e.g., Ebesutani et al., 2014; Taylor et al., 2007). Regarding the reliability of unit-weighted composite scores, we hypothesized that (H2a) a reliable general factor composite score would emerge, with comparatively weaker group factor composite scores. This hypothesis was extended (H2b) to the construct reliability and determinacy of the general and group factors. Regarding the extent to which the ASI-3 can lend itself to unidimensional modeling, we hypothesized that (H3a) the general factor would account for substantially more common variance than the group factors, and that it would (H3b) lend itself to unidimensional modeling without significant bias in general factor loadings. Following previous work (Ebesutani et al., 2014; Ghisi et al., 2016; Taylor et al., 2007), we also hypothesized that (H4) the ASI-3 would measure AS invariantly across gender. Because of the conflicting evidence in the literature (e.g., Ghisi et al., 2016; Sandin et al., 2007; Taylor et al., 2007), we made no hypothesis regarding gender differences. However, since women tend to report higher anxiety scores than men, we deemed it essential to control for anxiety. Furthermore, we hypothesized that (H5) the ASI-3 would moderately to strongly correlate with a measure of diffuse anxiety (convergent validity), and that (H6) this correlation would not be high enough to imply that the two constructs are identical (discriminant validity). Lastly, we hypothesized that (H7) the correlation between the ASI-3 and a measure of depression, originally significant and moderate or strong, would be significantly lowered after controlling for anxiety (divergent validity echoing the results of Ghisi et al., 2016).
Method
Procedure
This study is part of a larger research project aiming to progressively adapt measures of key constructs of the nomological network of the fundamental fears (Carleton, 2016) into Greek. It received ethical approval from the Institutional Review Board of the institution of the first and second authors. Before administering the survey, we informed all participants of the nature of the study and their right to withdraw at any given moment, in which case their data would be discarded. Furthermore, we informed participants that their data would be accessible only to scientists involved in the conduction and evaluation of this study (i.e., the present authors and reviewers).
Participants
We recruited a convenience sample representing the general Greek population through boosted Facebook posts between May and August 2022. We chose Facebook as the go-to platform because of its widespread use by individuals of varying ages. We did not use “smart-crowd” recruitment options to avoid bias, and we offered no reward for participation. In total, 913 individuals clicked on the survey link, 686 of whom began the survey. Of the 686, 30 did not complete the demographic questionnaire, 34 did not continue past the demographic questionnaire, and 1 participant abandoned the survey after the first questionnaire section. We excluded these 75 participants from the analysis, resulting in a data set with no missing data. The final sample thus consisted of 611 participants. The mean age of the sample was 32.75 (SD = 12.42, range = 18–74). Of the 611 participants, 293 (48%, Mage = 32.53, SDage = 11.74) identified as male and 318 (52%, Mage = 32.95, SDage = 13.03) as female.
Materials
Anxiety Sensitivity Index-3 (ASI-3)
The ASI-3 (Taylor et al., 2007) is an 18-item measure of AS that comprises three 6-item factors: (1) physical concerns (e.g., “It scares me when my heart beats rapidly.”), (2) cognitive concerns (e.g., “When I cannot keep my mind on a task, I worry that I might be going crazy.”), and (3) social concerns (e.g., “It is important for me not to appear nervous.”). The ASI-3 is scored on a 5-point Likert scale ranging from 0 (very little) to 4 (very much). In its initial development, the ASI-3 demonstrated excellent internal consistency as well as construct and criterion-related validity (Taylor et al., 2007). The present study uses a newly adapted Greek version of the ASI-3. The ASI-3 was translated into Greek following a forward-backward method. Three Greek psychologists with proficient knowledge of the English language and one bilingual psychologist generated two different sets of Greek (forward) and English (backward) translations. The investigators of this study compared the English and Greek versions produced by this process and observed minimal differences. Then, they compared the English versions to the original ASI-3, again observing minimal differences. Lastly, they reconciled the Greek versions, forming the version used in this study.
The Depression Anxiety Stress Scales-21 (DASS-21)
The DASS-21 is the abbreviated version of the DASS (Lovibond & Lovibond, 1993, 1995) and comprises three scales measuring depression, anxiety, and stress. The psychometric properties of the DASS-21 have been examined in diverse populations (e.g., Antony et al., 1998; Crawford et al., 2009; Norton, 2007; Osman et al., 2012). Together, these results suggest excellent reliability and validity. The DASS was adapted to Greek by Lyrakos et al. (2011) and has demonstrated comparable reliability and validity. The psychometric properties of the Greek abbreviated version were examined by Pezirkianidis et al. (2018) and found to be good. We used only the anxiety and depression subscales in this study.
Data Analysis
We conducted data analysis using SPSS v. 26 (IBM Corp, 2019) and R (R Core Team, 2021), employing the following R packages: “foreign” (0.8-84; R Core Team, 2022), “lavaan” (v0.6-12; Rosseel, 2012), “psych” (v2.2.9; Revelle, 2022), “EFAtools” (v0.4.4; Steiner, 2023), “EGAnet” (v1.1.1, Golino & Christensen, 2022), “semTools” (v0.5-6; Jorgensen et al., 2020), “ggpubr” (v0.6.0; Kassambara, 2023a), “rstatix” (v0.7.2; Kassambara, 2023b), “car” (v3.1-2, Fox et al., 2023), “responsePatterns (v0.1.0; Rihacek & Gotfried, 2022), and “emmeans” (v1.8.6; Lenth, 2023).
Initially, we employed an autocorrelation method for identifying patterned responses (i.e., 1–2–3–4–5–1–2–3–4–5) as described by Gottfried et al. (2022). We manually examined the participants with the highest autocorrelation coefficients and did not identify any suspicious cases. Before examining the ASI-3’s factor structure, we identified multivariate outliers by examining the χ2 distribution of the Mahalanobis distances of the ASI-3 items. A suggested threshold for outlier detection is a p-value of .001 (Tabachnick & Fidell, 2018). We then ran preliminary analyses, including calculating descriptive statistics and interitem correlations, thereby assessing multivariate normality. Afterward, we took a two-step analytic approach to examining the factor structure of the ASI-3. We first conducted exploratory factor analysis (EFA) to identify the simplest latent structure of the ASI-3 and assess the presence of cross-loadings. Then, we used confirmatory factor analyses (CFA) to compare alternative models. Before fitting restricted confirmatory models, evaluating the presence of substantial cross-loadings is important as they can bias parameter estimates, especially in bifactor models (Reise et al., 2010).
Factor Analyses (H1)
We conducted EFAs using the “psych” package, using oblimin rotation in all EFAs, and, following the recommendations for EFA on ordinal data (Goretzko et al., 2019), the estimator and input were weighted least squares (WLS) and a polychoric correlation matrix, respectively. Factor retention was guided by exploratory graph analysis (EGA; Golino & Epskamp, 2017) using the “EGAnet” package. EGA applies a network model to model the covariance structure and has the added benefit of identifying the items belonging to each retrieved community (i.e., dimension). It has been shown to outperform traditional factor retention techniques (Golino et al., 2020; Golino & Epskamp, 2017; Cosemans et al., 2022). Considering the results of the EGA, we conducted two EFAs: a hierarchical one and a bifactor one using the Schmid-Leiman orthogonalization (SL; Giordano & Waller, 2020; Schmid & Leiman, 1957).
We conducted CFAs using the “lavaan” package. Because of the ordinal nature of the data, we used the WLSMV estimator, which implies the diagonally weighted least squares (DWLS) estimator with adjusted means and variances and robust standard errors. Li’s (2015, 2016) simulation studies support using WLSMV and DWLS over the more traditional maximum likelihood (ML) estimation when data are ordinal. After previous work, we considered five alternative models (Ebesutani et al., 2014; Ghisi et al., 2016; Taylor et al., 2007). The first model was a unidimensional model with all items loading on a general AS factor. The second and third models were correlated factor models, with two factors each representing cognitive-physical concerns + social concerns and cognitive-social concerns + physical concerns solutions, respectively. The fourth model was a correlated three-factor model. The fifth model was a bifactor model with one general factor and three orthogonal group factors. We evaluated model fit based on several fit indexes. These include the χ2 test, the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). Assessment of these fit indexes was as follows: The χ2 test should not be significant, CFI should be greater than .90/.95 (indicating adequate and good fit, respectively), RMSEA should be lower than .08/.05 (indicating adequate and good fit, respectively), and SRMR should be lower than .10/.05 (indicating adequate and good fit, respectively; Tabachnick & Fidell, 2018). Because the χ2 test is oversensitive to sample size (Bentler & Bonett, 1980; Saris et al., 2009), we paid attention mainly to the latter three indexes. We then compared the models based on χ2 difference tests (Satorra & Bentler, 2001) with a mean and variance-adjusted test statistic (see Satorra & Bentler, 1994; Satterthwaite, 1941). We did not include models that failed to evidence adequate fit in this analysis. Like Ghisi et al. (2016), we tested no hierarchical models as they evidence a fit identical to their correlated-factor counterparts (Brown, 2015).
Reliability and Dimensionality (H2a,b; H3a,b)
To examine the dimensionality and reliability of the ASI-3, we computed the indices suggested by Rodriguez et al. (2016a, 2016b) based on the confirmatory bifactor solution. These include the omega total (ωt), omega hierarchical (ωh), omega subscale (ωs), and omega hierarchical subscale (ωhs) congeneric reliability coefficients (McDonald, 1999; Reise et al., 2013; Zinbarg et al., 2006); the FD (factor determinacy; Gorsuch, 1980; Grice, 2001) and H coefficients (construct replicability; Hancock & Mueller, 2001); the ECV (explained common variance because of a general factor; Sijtsma, 2009; Stucky & Edelen, 2014; ten Berge & Sočan, 2004) and the PUC (percentage of uncontaminated correlations; Bonifay et al., 2015; Reise et al, 2013). For a detailed review and application of these coefficients, see the relevant substantive literature (Rodriguez et al., 2016a, 2016b).
Measurement Invariance Across Gender (H4)
We examined measurement invariance across gender in both the correlated three-factor and bifactor solutions following the process described by D’Urso et al. (2021) for categorical variables. First, we conducted CFAs for each gender independently and then conducted an unconstrained multigroup CFA (MG-CFA) to assess configural invariance. Because of the ordinal nature of the data, we assessed metric and scalar invariance simultaneously by applying equality constraints on loadings and thresholds instead of progressively applying them on loadings and intercepts (as is the case with continuous data). Accordingly, we compared the configural and constrained models based on differences in CFI (ΔCFI; Chen, 2007) and a robust χ2 difference test with a mean and variance-adjusted test statistic. The alternative hypothesis of noninvariance is discarded if the constrained model shows small changes in CFI (i.e., ΔCFI < .01) and if the χ2 difference test is not significant. We paid attention mainly to ΔCFI as it is not affected by sample size and model complexity as much as the χ2 difference test (Chen, 2007; Cheung & Rensvold, 2002). Following the recommendations of Sass et al. (2014), we compared the results to those of a parallel measurement invariance examination using models estimated through robust maximum likelihood estimation (MLR). Given the results, we conducted an ANCOVA examining differences in unit-weighted ASI-3 total scores between men and women after accounting for diffuse anxiety.
Convergent, Discriminant, and Divergent Validity (H5; H6; H7)
We examined convergent, discriminant, and divergent validity by exploring the strength of ASI-3’s relationship with the DASS-21 anxiety and depression subscales. We established the relationships between the latent constructs as zero-order correlations estimated through CFA. Before estimating the model, we identified and excluded multivariate outliers for the combined set of ASI-3 and DAS-21 items. We tested discriminant validity with the χ2 (sys) method, as described by Rönkkö and Cho (2020). We examined divergent validity by computing partial correlations based on these zero-order correlations.
Results
Table 1 shows the descriptive statistics and Cronbach’s α for the unit-weighted composite scores of the ASI-3 and the DASS-21. The mean interitem correlation of the ASI-3 items was .45. The assumption of multivariate normality did not hold, as evidenced by a statistically significant Mardia’s test: b1,p = 30.22, p < .001; b2,p = 423.5, p < .001.
Exploratory Factor Analysis
The overall KMO value was .93, and Bartlett’s test of sphericity was significant at the .001 level. The EGA identified three communities, corresponding exactly to the three-factor solution proposed by Taylor et al. (2007), with a total correlation of .77 and an average entropy of −3.19. The estimated network may be seen in Figure 1. Thereafter, in each of the EFAs, we retained three lower-order factors. The results of the hierarchical EFA revealed that no cross-loading over .21 existed in the first order. Furthermore, all items loaded above .30 on their corresponding factor, with most loading higher than .60. Regarding the second-order solution, all first-order factors loaded strongly on the general AS factor. We observed the same pattern of results in the bifactor EFA. All items loaded above .30 on the general AS and their corresponding group factors except for item 17 (“I think it would be horrible for me to faint in public.”), which loaded at .025 on its corresponding group factor (i.e., social concerns). Like the bifactor EFA results of Ebesutani et al. (2014), the average drop in factor loadings for the cognitive concerns subscale was .43, after controlling for the general AS factor. The average drops for the social and physical concerns subscales were comparatively much smaller, both standing at .22. Factor loadings for both EFAs are shown in Table 2.
Confirmatory Factor Analysis (H1)
Table 3 shows the fit indices for all models. Both the bifactor model (χ2(117) = 296.89, CFI = .982, RMSEA = .052, SRMR = .033) and the correlated three-factor model (χ2(132) = 397.04, CFI = .974, RMSEA = .059, SRMR = .043) showed strong fit, while the rest of the models failed to show adequate fit. As hypothesized, the χ2 difference test revealed that the bifactor model fit the data significantly better than the correlated three-factor model; χ2diff(15) = 78, p < .001. Table 4 shows the factor loadings for both models, where all items loaded strongly on their corresponding factors, except for items 17, 2 (“When I cannot keep my mind on a task, I worry that I might be going crazy.”), and 5 (“It scares me when I am unable to keep my mind on a task.”), which failed to load above .30 on their corresponding group factor in the bifactor model. A similar pattern of factor loading drops occurred as in the EFA, with the cognitive concerns items dropping an average of .45, the physical concerns items dropping an average of .28, and the social concerns items dropping an average of .27.
Reliability and Dimensionality (H2a, H2b; H3a, H3b)
The ωt and ωh coefficients for the ASI-3 total score were .96 and .86, respectively, indicating that 89% of the variance in the ASI-3 total score is attributable to the general AS factor. Regarding the subscale composite scores, 36% of the variance was attributable to the group factor for the social concerns dimension (ωs = .86, ωhs = .31), 42% for the physical concerns dimensions (ωs = .92, ωhs = .39), and 18% for the cognitive concerns dimension (ωs = .93, ωhs = .17). The FD coefficient was .93 for the general AS factor, .81 for social concerns, .87 for physical concerns, and .66 for cognitive concerns. Moreover, the H coefficient was .86 for the general AS factor, .65 for social, .76 for physical, and .44 for cognitive concerns. Together, these results point to a reliable general factor across measurement methods and to group factors that, after accounting for the general factor, mostly fail to reach acceptable thresholds (H2a, H2b). An exception is the physical concerns subscale, whose values trended toward being acceptable. Furthermore, the ECV and PUC were .67 and .71, respectively, indicating that (H3a) the general factor accounts for substantially more common variance than the group factors, and that the ASI-3 may lend itself to unidimensional modeling (H3b). However, a calculation of the differences in general factor loadings between the bifactor and the unidimensional models yielded an average bias of .14, which, to our estimation, is high enough to preclude any such action.
Measurement Invariance and Gender Differences (H4)
Table 5 shows the fit indices of all models considered in the measurement invariance examination. Both the bifactor and correlated three-factor models showed a good fit when investigated separately in males (n = 273) and females (n = 301). Configural invariance was established, as evidenced by the fit indices of the unconstrained MG-CFA models (correlated three-factor model: CFI = .971, RMSEA = .061, SRMR = .054; bifactor model: CFI = .983, RMSEA = .049, SRMR = .043). Metric and scalar invariances were also supported, as evidenced by a trivial difference in model fit when comparing the unconstrained and constrained MG-CFA models (ΔCFI < .01 in both the bifactor and correlated three-factor model comparisons). Furthermore, the results of the parallel measurement invariance examination using the MLR estimator replicated these findings. However, note that, while estimating the bifactor model using MLR, some estimated variances were negative, resulting in the collapse of the cognitive concerns dimension. To our best knowledge, this issue arises from modeling ordinal data with the MLR estimator.
Regarding the ANCOVA, the assumptions of linearity, homogeneity of regression slopes, normality of the residuals, and homogeneity of variances held. The raw difference in AS scores between men (M = 19.6) and women (M = 27.8) was 8.2. The results revealed a statistically significant relationship between diffuse anxiety and AS F(1, 571) = 490.41, p < .05, η2G = .46, as well as a statistically significant effect of gender on AS after controlling for diffuse anxiety, F (1, 571) = 19.5, p < .05, η2G = .03. An examination of the estimated marginal means revealed an adjusted difference of 4.42, p < .001 in AS scores.
Convergent, Discriminant, and Divergent Validity (H5; H6; H7)
The combined measurement model of the ASI-3 and the DASS-21 anxiety and depression subscales demonstrated adequate fit, χ2(458) = 1236.12, CFI = .952, RMSEA = .057, SRMR = .06. As hypothesized (H5), the correlation between the AS general factor and anxiety was strong and statistically significant (r = .80, LCI = .74, UCI = .85, p < .001). Given that the UCI value was greater than .80, we compared this model with a model in which the correlation between the AS general factor and anxiety was fixed to .90. The χ2 difference test was significant (χ2diff(1) = 15.07, p < .001), allowing for the classification of the discriminant validity problem as marginal. As hypothesized (H6), this finding indicates that the two measures most probably measure distinct constructs. Lastly, the correlation between the general AS factor and depression (H7) was strong and statistically significant, r = .62, LCI = .55, UCI = .68, p < .001. After controlling for anxiety, it fell to a comparatively much weaker value, r = .1, LCI = .01, UCI = .18, p < .05.
Discussion
The present study examined the psychometric properties of a newly translated Greek version of the ASI-3. We followed a two-step analytic procedure: First, EGA and EFA identified a three-factor structure with mostly minimal cross-loadings; then, CFAs comparing a series of nested models revealed that a bifactor model fit the data best, followed by a correlated three-factor model. Examining the reliability and dimensionality of the ASI-3 using the confirmatory bifactor model revealed a strong general factor and comparatively weaker group factors. These results largely align with previous investigations of the psychometric properties of the ASI-3 (Ebesutani et al., 2014; Ghisi et al., 2016; Rodriguez et al., 2016b). Although the bifactor model fit the data best, a direct interpretation of its substantive meaning for the structure of the underlying trait should be avoided. Valid concerns for its tendency to outperform hierarchical models in terms of fit indices have been previously raised, and good fit is not sufficient to identify the structure of psychopathology as it works on a neurobiological level (see Bonifay et al., 2016). To this end, we echo the sentiment of Ebesutani et al. (2014) and Reise et al. (2010) regarding the importance of neurobiological evidence.
Nevertheless, these results suggest that the ASI-3 operates in largely similar ways across cultures. However, we did find subtle differences between the present investigation and the previous investigations on U.S. samples (Ebesutani et al., 2014; Osman et al., 2010). Specifically, item 2 (“When I cannot keep my mind on a task, I worry that I might be going crazy.”) and item 5 (“It scares me when I am unable to keep my mind on a task.”) failed to adequately load on their respective group factor (i.e., cognitive concerns) in the confirmatory bifactor solution. This finding is partly in line with the results reported by Ghisi et al. (2016), gathered from an Italian sample, according to which only item 2 and item 4 (“It scares me when I am unable to keep my mind on a task.”) loaded adequately on their respective group factor. The large loading differences observed between the correlated three-factor model and the bifactor model in the cognitive concerns dimension are tied to this. These findings suggest that the cognitive concerns dimension may be a purer depiction of generalized AS in southern European cultures. In our investigation, however, item 17 (“I think it would be horrible for me to faint in public.”) also failed to load adequately on its respective group factor (i.e., social concerns) after controlling for the general AS factor. This finding replicates the previously mentioned concerns regarding the manifestation of domain-specific AS across cultures.
The results about the reliability and dimensionality of the ASI-3 suggest that, when it is used in applied clinical or research settings, the most reliable and consistent scores are those of the general factor. Specifically, scores reflecting the general AS factor were reliable across measurement methods, as evidenced by its high ωh, FD, and H values. Conversely, unit-weighted composite scores for the group factors did not reflect sufficient reliable variance when we controlled for the general factor and are, thus, not particularly informative. Moreover, group factors were not sufficiently determined by their corresponding items, as evidenced by their subthreshold FD and H values. Consequently, the present results do not support using domain-specific AS in applied clinical or research contexts. However, the physical concerns dimension accounted for a substantial amount of variance in its corresponding unit-weighted composite score (42%) after we controlled for the general factor as well as showing FD (.87) and H (.76) coefficients that were close to the proposed thresholds. We also found these results in the slightly larger factor loadings that the social concerns items had on their respective group factors compared to the rest of the items. This finding suggests that physical concerns are more salient and determined in the Greek population, possibly because of a greater tendency of the Greek population for somatization rather than for fearing loss of mental control or for externalizing anxiety to socially relevant objects. There are previous reports that somatization is more prevalent in the Greek than in other cultures (Kontoangelos et al., 2015), but, to our estimation, we need more findings to support this contention. Moreover, the general AS factor accounted for 67% of the common variance in the ASI-3 items, suggesting that common variance on the item level is largely attributable to it. However, comparing the general factor loadings between the unidimensional and bifactor models revealed a relatively high distortion, suggesting that unidimensional modeling might not be plausible for the ASI-3. This conclusion agrees with the bad fit evidenced by the present unidimensional solution and previous investigations (e.g., Ebesutani et al., 2014).
Another goal of the present investigation was to investigate whether the ASI-3 measures AS invariantly across gender. This investigation revealed that configural, metric, and scalar invariances held in the three-factor and bifactor models. We share the limitation of this analysis with Ebesutani et al. (2014), in that metric and scalar invariance models were estimated simultaneously with fixed loadings and thresholds, rather than independently with fixed loadings and intercepts. This analysis is more appropriate because of the ordinal nature of the data but departs from the original measurement invariance examination conducted by Taylor et al. (2007). However, a parallel examination treating the data as continuous using the MLR estimator revealed similar results. After we controlled for diffuse anxiety, the ANCOVA revealed a weak, statistically significant difference in AS scores between men and women.
The last aim of the present investigation was to examine the convergent, discriminant, and divergent validity of the ASI-3. In line with previous theory and research (Ghisi et al., 2016; McNally, 1996; Reiss, 1997), the results revealed that the ASI-3 is strongly related with, and sufficiently distinct from, diffuse anxiety, and that its relationship with anxiety is markedly stronger than its relationship with depression. These results corroborate the standing of AS as an important transdiagnostic construct that, however, is more specific to anxiety psychopathology.
Limitations
The present study is not without its limitations. We included no clinical samples in the analysis, so we cannot generalize the present results to clinical populations. Furthermore, following the results of D’Urso et al. (2021), ΔCFI and χ2 tests are not appropriate when using the WLSMV estimator. Modeling ordinal data as continuous is not ideal, either, so we cannot be certain whether the present results about measurement invariance are conclusive. We need future methodological advances to inform measurement invariance examinations of ordinal data. Nevertheless, these results do suggest that, to the best of our knowledge, one can safely compare AS scores between genders. Furthermore, the present validity investigation is limited in that it included no other explanatory constructs from the nomological network of AS and should thus be considered preliminary. Key constructs to be considered are intolerance of uncertainty, the fear of negative evaluation, illness/injury sensitivity, and the fear of pain (see Carleton et al., 2014, for a comprehensive investigation).
Conclusion
This study represents the first examination of the psychometric properties of the ASI-3 in a Greek population and paves the way for related constructs to be adapted and become available for cross-cultural studies. The present results reveal that the Greek adaptation of the ASI-3 behaves largely like the original version. Specifically, the results provide cross-cultural evidence in favor of the three-factor structure of the ASI-3 as well as for the presence of a strong, reliable general AS factor relative to weaker domain-specific factors (Ebesutani et al., 2014; Ghisi et al., 2016; Osman et al., 2010; Rodriguez et al., 2016b). Furthermore, the present results suggest that the Greek ASI-3 measures AS invariantly across gender, and that women report higher AS scores than men. Lastly, preliminary evidence on the Greek ASI-3 s convergent, discriminant, and divergent validity is demonstrated through its associations with the DASS-21 anxiety and depression subscales. The results suggest that the Greek ASI-3 has adequate psychometric properties and is appropriate for empirical research. Future studies should explore its criterion-related validity by administering the Greek ASI-3 to clinical populations and its relationship to other key constructs of AS’s nomological network.
References
(2015). Support for the general and specific bifactor model factors of anxiety sensitivity. Personality and Individual Differences, 74, 78–83. https://doi.org/10.1016/j.paid.2014.10.003
(2014). Unique relations among anxiety sensitivity factors and anxiety, depression, and suicidal ideation. Journal of Anxiety Disorders, 28(2), 266–275. https://doi.org/10.1016/j.janxdis.2013.12.004
(2018). Lower-order anxiety sensitivity and intolerance of uncertainty dimensions operate as specific vulnerabilities for social anxiety and depression within a hierarchical model. Journal of Anxiety Disorders, 53, 91–99. https://doi.org/10.1016/j.janxdis.2017.08.002
(1998). Psychometric properties of the 42-item and 21-item versions of the Depression Anxiety Stress Scales in clinical groups and a community sample. Psychological Assessment, 10(2), 176–181. https://doi.org/10.1037/1040-3590.10.2.176
(2002). Gender differences in anxiety: An investigation of the symptoms, cognitions, and sensitivity toward anxiety in a nonclinical population. Behavioural and Cognitive Psychotherapy, 30(2), 227–231. https://doi.org/10.1017/s1352465802002114
(2006). Confirmatory factor analysis and psychometric properties of the Anxiety Sensitivity Index – Revised in clinical and normative populations. European Journal of Psychological Assessment, 22(2), 116–126. https://doi.org/10.1027/1015-5759.22.2.116
(2009). The Anxiety Sensitivity Index – Revised: Confirmatory factor analysis, structural invariance in Caucasian and African American samples, and score reliability and validity. Assessment, 16(2), 165–180. https://doi.org/10.1177/1073191108328809
(1980). Significance tests and goodness of fit in the analysis of covariance structures. Psychological Bulletin, 88, 588–606. https://doi.org/10.1037/0033-2909.88.3.588
(2007). Anxiety sensitivity: Selective review of promising research and future directions. Expert Review of Neurotherapeutics, 7(2), 97–101. https://doi.org/10.1586/14737175.7.2.97
(2017). Anxiety sensitivity as a predictor of outcome in the treatment of obsessive-compulsive disorder. Journal of Behavior Therapy and Experimental Psychiatry, 57, 113–117. https://doi.org/10.1016/j.jbtep.2017.05.003
(2015). When are multidimensional data unidimensional enough for structural equation modeling? An evaluation of the DETECT Multidimensionality Index. Structural Equation Modeling: A Multidisciplinary Journal, 22(4), 504–516. https://doi.org/10.1080/10705511.2014.938596
(2016). Three concerns with applying a bifactor model as a structure of psychopathology. Clinical Psychological Science, 5(1), 184–186. https://doi.org/10.1177/2167702616657069
(2015). Confirmatory factor analysis for applied research (2nd ed.). Guilford Press.
(2018). Reliability, validation and norms of the Chinese version of Anxiety Sensitivity Index 3 in a sample of military personnel. PLoS One, 13(8), Article
e0201778 . https://doi.org/10.1371/journal.pone.0201778(2016). Fear of the unknown: One fear to rule them all? Journal of Anxiety Disorders, 41, 5–21. https://doi.org/10.1016/j.janxdis.2016.03.011
(2014). Revisiting the fundamental fears: Toward establishing construct independence. Personality and Individual Differences, 63, 94–99. https://doi.org/10.1016/j.paid.2014.01.040
(2013). Sensitivity and specificity of a brief personality screening instrument in predicting future substance use, emotional, and behavioral problems: 18-month predictive validity of the Substance Use Risk Profile scale. Alcoholism: Clinical and Experimental Research, 37, E281–E290. https://doi.org/10.1111/j.1530-0277.2012.01931.x
(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464–504. https://doi.org/10.1080/10705510701301834
(2002). Evaluating goodness-of-fit indexes for testing measurement invariance. Structural Equation Modeling, 9(2), 233–255. https://doi.org/10.1207/s15328007sem0902_5
(2001). What’s the use of neuroticism? Personality and Individual Differences, 31(3), 383–400. https://doi.org/10.1016/s0191-8869(00)00144-6
(2022). Exploratory graph analysis for factor retention: Simulation results for continuous and binary data. Educational and Psychological Measurement, 82(5), 880–910. https://doi.org/10.1177/00131644211059089
(2009). A convenient method of obtaining percentile norms and accompanying interval estimates for self-report mood scales (DASS, DASS-21, HADS, PANAS, and SAD). British Journal of Clinical Psychology, 48(2), 163–180. https://doi.org/10.1348/014466508x377757
(2003). The Anxiety Sensitivity Index – Revised: Psychometric properties and factor structure in two nonclinical samples. Behaviour Research and Therapy, 12, 1427–1449. https://doi.org/10.1016/S0005-7967(03)00065-2
(2002). The Anxiety Sensitivity Index for Children: Factor structure and relation to panic symptoms in an adolescent sample. Behaviour Research and Therapy, 40(7), 839–852. https://doi.org/10.1016/S0005-7967(01)00076-6
(2021). Scale length does matter: Recommendations for measurement invariance testing with categorical factor analysis and item response theory approaches. Behavior Research Methods, 54(5), 1–32. https://doi.org/10.3758/s13428-021-01690-7
(2014). A bifactor model of anxiety sensitivity: Analysis of the Anxiety Sensitivity Index-3. Journal of Psychopathology and Behavioral Assessment, 36(3), 452–464. https://doi.org/10.1007/s10862-013-9400-3
(2009). Attributional style and anxiety sensitivity as maintenance factors of posttraumatic stress symptoms: A prospective examination of a diathesis–stress model. Journal of Behavior Therapy and Experimental Psychiatry, 40(4), 544–557. https://doi.org/10.1016/j.jbtep.2009.07.005
(1982). Fear of anxiety: Development and validation of an assessment scale. University of Illinois Press.
(2008). Anxiety sensitivity factor structure among Brazilian patients with anxiety disorders. Journal of Psychopathology and Behavioral Assessment, 31(3), 246–255. https://doi.org/10.1007/s10862-008-9103-3
(2015). Evaluation of the Anxiety Sensitivity Index-3 among treatment-seeking smokers. Psychological Assessment, 27(3), 1123–1128. https://doi.org/10.1037/pas0000112
(2007). Anxiety sensitivity as a moderator of the relation between trauma exposure frequency and posttraumatic stress symptomatology. Journal of Cognitive Psychotherapy, 20(2), 201–213. https://doi.org/10.1891/jcop.20.2.201
(2013). Anxiety sensitivity and intolerance of uncertainty: Evidence of incremental specificity in relation to health anxiety. Personality and Individual Differences, 55(6), 640–644. https://doi.org/10.1016/j.paid.2013.05.016
(2014). How do elements of a reduced capacity to withstand uncertainty relate to the severity of health anxiety? Cognitive Behaviour Therapy, 43(3), 262–274. https://doi.org/10.1080/16506073.2014.929170
(2005). Anxiety sensitivity and worry. Personality and Individual Differences, 38(5), 1223–1229. https://doi.org/10.1016/j.paid.2004.08.005
(2023). Companion to applied regression. https://cran.r-project.org/web/packages/car/car.pdf
(1895). Obsessions and phobias: Their psychical mechanisms and their etiology. S. Freud (1924), Collected papers (Vol. 1). Hograth Press.
(2016). Factor structure and psychometric properties of the Anxiety Sensitivity Index-3 in an Italian community sample. Frontiers in Psychology, 7, 1–13. https://doi.org/10.3389/fpsyg.2016.00160
(2020). Recovering bifactor models: A comparison of seven methods. Psychological Methods, 25(2), 143–156. https://doi.org/10.1037/met0000227
(2022). Autocorrelation screening: A potentially efficient method for detecting repetitive response patterns in questionnaire data. Practical Assessment, Research, and Evaluation, 27(2), 1–11. https://doi.org/10.7275/vyxb-gt24
(2017). Exploratory graph analysis: A new approach for estimating the number of dimensions in psychological research. PLoS One, 12, Article
e0174035 . https://doi.org/10.1371/journal.pone.0174035(2022). EGAnet: Exploratory graph analysis – A framework for estimating the number of dimensions in multivariate data using network psychometrics. R package version 1.1.1. https://CRAN.R-project.org/package=EGAnet
(2020). Investigating the performance of exploratory graph analysis and traditional techniques to identify the number of latent factors: A simulation and tutorial. Psychological Methods, 25(3), 292–320. https://doi.org/10.1037/met0000255
(2019). Exploratory factor analysis: Current use, methodological developments and recommendations for good practice. Current Psychology, 40(7), 3510–3521. https://doi.org/10.1007/s12144-019-00300-2
(1980). Factor score reliabilities and domain validities. Educational and Psychological Measurement, 40(4), 895–897. https://doi.org/10.1177/001316448004000412
(2001). Computing and evaluating factor scores. Psychological Methods, 6, 430–450. https://doi.org/10.1037/1082-989X.6.4.430
(2001).
Rethinking construct reliability within latent variable systems . In R. CudeckS. du ToitD. SörbomEds., Structural equation modeling: Present and future – A Festschrift in honor of Karl Jöreskog (pp. 195–216). Scientific Software International.(2022). Psychometric properties of the Anxiety Sensitivity Index-3 in adults with substance use disorders. Journal of Substance Abuse Treatment, 132, Article
108507 . https://doi.org/10.1016/j.jsat.2021.108507. (2019). IBM SPSS Statistics for Windows, Version 26.0.
(1999). Gender differences in the etiology of anxiety sensitivity: A twin study. Journal of Gender Specific Medicine, 2(2), 39–44.
(2018). Toward a greater understanding of anxiety sensitivity across groups: The construct validity of the Anxiety Sensitivity Index-3. Psychiatry Research, 268, 72–81. https://doi.org/10.1016/j.psychres.2018.07.007
(2020). semTools: Useful tools for structural equation modeling, R package version 0.5-6. https://CRAN.R-project.org/package=semTools
(2023a). “ggplot2” based publication ready plots. https://cran.r-project.org/web/packages/ggpubr/ggpubr.pdf
(2023b). Pipe-friendly framework for basic statistical tests. https://cran.r-project.org/web/packages/rstatix/rstatix.pdf
(2012). Construct validity of the Anxiety Sensitivity Index-3 in clinical samples. Assessment, 19(1), 89–100. https://doi.org/10.1177/1073191111429389
(2017). White matter correlates of anxiety sensitivity in panic disorder. Journal of Affective Disorders, 207, 148–156. https://doi.org/10.1016/j.jad.2016.08.043
(2016). Anxiety sensitivity and its factors in relation to generalized anxiety disorder among adolescents. Journal of Abnormal Child Psychology, 44(2), 233–244. https://doi.org/10.1007/s10802-015-9991-0
(2015). Greek college students and psychopathology: New insights. International Journal of Environmental Research and Public Health, 12(5), 4709–4725. https://doi.org/10.3390/ijerph120504709
(2015). Distress tolerance in OCD and anxiety disorders, and its relationship with anxiety sensitivity and intolerance of uncertainty. Journal of Anxiety Disorders, 33, 8–14. https://doi.org/10.1016/j.janxdis.2015.04.003
(2023). Estimated marginal means, aka least-squares means. https://cran.r-project.org/web/packages/emmeans/emmeans.pdf
(1998). Gender differences in anxiety disorders and anxiety symptoms in adolescents. Journal of Abnormal Psychology, 107(1), 109–117. https://doi.org/10.1037/0021-843X.107.1.109
(2015). Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods, 48(3), 936–949. https://doi.org/10.3758/s13428-015-0619-7
(2016). The performance of ML, DWLS, and ULS estimation with robust corrections in structural equation models with ordinal variables. Psychological Methods, 21(3), 369–387. https://doi.org/10.1037/met0000093
(2012). Korean Anxiety Sensitivity Index-3: Its factor structure, reliability, and validity in nonclinical samples. Psychiatry Investigation, 9(1), 45–53. https://doi.org/10.4306/pi.2012.9.1.45
(1995). The structure of negative emotional states: Comparison of the Depression Anxiety Stress Scales (DASS) with the Beck Depression and Anxiety Inventories. Behaviour Research and Therapy, 33, 335–342. https://doi.org/10.1016/0005-7967(94)00075-U
(1993). Manual for the Depression Anxiety Stress Scales (DASS). Psychology Foundation.
(2011). Translation and validation study of the depression anxiety stress scale in the Greek general population and in a psychiatric patient’s sample. European Psychiatry, 26(S2), 1731. https://doi.org/10.1016/S0924-9338(11)73435-6
(1992). Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety Disorders, 6(3), 241–247. https://doi.org/10.1016/0887-6185(92)90036-7
(1999). Test theory: A unified approach. Erlbaum.
(2009). Stressful life events, anxiety sensitivity, and internalizing symptoms in adolescents. Journal of Abnormal Psychology, 118(3), 659–669. https://doi.org/10.1037/a0016499
(2011). Gender differences in anxiety disorders: Prevalence, course of illness, comorbidity and burden of illness. Journal of Psychiatric Research, 45(8), 1027–1035. https://doi.org/10.1016/j.jpsychires.2011.03.006
(2002). Anxiety sensitivity and panic disorder. Biological Psychiatry, 52(10), 938–946. https://doi.org/10.1016/s0006-3223(02)01475-0
(1996).
Anxiety sensitivity is distinguishable from trait anxiety . In R. M. RapeeEd., Current controversies in the anxiety disorders (pp. 214–227). Guilford Press.(2015). Anxiety sensitivity and the anticipation of predictable and unpredictable threat: Evidence from the startle response and event-related potentials. Journal of Anxiety Disorders, 33, 62–71. https://doi.org/10.1016/j.janxdis.2015.05.003
(2007). Depression Anxiety and Stress Scales (DASS-21): Psychometric analysis across four racial groups. Anxiety, Stress & Coping, 20(3), 253–265. https://doi.org/10.1080/10615800701309279
(2014). Anxiety Sensitivity Index (ASI-3) subscales predict unique variance in anxiety and depressive symptoms. Journal of Anxiety Disorders, 28(2), 115–124. https://doi.org/10.1016/j.janxdis.2013.04.009
(2004). Neuroticism: A non-informative marker of vulnerability to psychopathology. Social Psychiatry and Psychiatric Epidemiology, 39(11), 906–912. https://doi.org/10.1007/s00127-004-0873-y
(2010). The Anxiety Sensitivity Index-3: Analyses of dimensions, reliability estimates, and correlates in nonclinical samples. Journal of Personality Assessment, 92, 45–52. https://doi.org/10.1080/00223890903379332
(2012). The Depression Anxiety Stress Scales-21 (DASS-21): Further examination of dimensions, scale reliability, and correlates. Journal of Clinical Psychology, 68, 1322–1338. https://doi.org/10.1002/jclp.21908
(2014). Direct and indirect predictors of social anxiety: The role of anxiety sensitivity, behavioral inhibition, experiential avoidance and self-consciousness. Comprehensive Psychiatry, 55(8), 1875–1882. https://doi.org/10.1016/j.comppsych.2014.08.045
(1987). The Anxiety Sensitivity Index: Construct validity and factor analytic structure. Journal of Anxiety Disorders, 2, 117–121. https://doi.org/10.1016/0887-6185(87)90002-8
(2018). Psychometric properties of the Depression, Anxiety, Stress Scales-21 (DASS-21) in a Greek sample. Psychology, 9, 2933–2950. https://doi.org/10.4236/psych.2018.915170
(2015). Neural correlates of anxiety sensitivity in panic disorder: A functional magnetic resonance imaging study. Psychiatry Research: Neuroimaging, 233(2), 95–101. https://doi.org/10.1016/j.pscychresns.2015.05.013
(1991). Neo-conditioning and the classical theory of fear acquisition. Clinical Psychology Review, 11(2), 155–173. https://doi.org/10.1016/0272-7358(91)90093-a
(2000). Psychological and physiological predictors of response to carbon dioxide challenge in individuals with panic disorder. Journal of Abnormal Psychology, 109(4), 616–623. https://doi.org/10.1037/0021-843X.109.4.616
. (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/
. (2022). Package “foreign”. https://cran.r-project.org/web/packages/foreign/foreign.pdf
(1991). Expectancy model of fear, anxiety, and panic. Clinical Psychology Review, 11(2), 141–153. https://doi.org/10.1016/0272-7358(91)90092-9
(1997). Trait anxiety: It’s not what you think it is. Journal of Anxiety Disorders, 11(2), 201–214. https://doi.org/10.1016/s0887-6185(97)00006-6
(1985).
The expectancy model of fear . In S. ReissR. R. BootzinEds., Theoretical issues in behavior therapy (pp. 107–122). Academic Press.(1988). Anxiety sensitivity, injury sensitivity, and individual differences in fearfulness. Behaviour Research and Therapy, 26(4), 341–345. https://doi.org/10.1016/0005-7967(88)90088-5
(1986). Anxiety sensitivity, anxiety frequency, and the predictions of fearfulness. Behaviour Research and Therapy, 24, 341–345. https://doi.org/10.1016/0005-7967(86)90143-9
(2010). Bifactor models and rotations: Exploring the extent to which multidimensional data yield univocal scale scores. Journal of Personality Assessment, 92(6), 544–559. https://doi.org/10.1080/00223891.2010.496477
(2007). The role of the bifactor model in resolving dimensionality issues in health outcomes measures. Quality of Life Research, 16(S1), 19–31. https://doi.org/10.1007/s11136-007-9183-7
(2013). Multidimensionality and structural coefficient bias in structural equation modeling. Educational and Psychological Measurement, 73(1), 5–26. https://doi.org/10.1177/0013164412449831
(2022). psych: Procedures for Psychological, Psychometric, and Personality Research (R package version 2.2.9). Northwestern University. https://CRAN.R-project.org/package=psych
(2015). Psychometric properties of the Anxiety Sensitivity Index-3 in an acute and heterogeneous treatment sample. Journal of Anxiety Disorders, 36, 99–102. https://doi.org/10.1016/j.janxdis.2015.09.010
(2022). Screening for careless responding patterns. https://cran.r-project.org/web/packages/responsePatterns/index.html
(2016a). Evaluating bifactor models: Calculating and interpreting statistical indices. Psychological Methods, 21(2), 137–150. https://doi.org/10.1037/met0000045
(2016b). Applying bifactor statistical indices in the evaluation of psychological measures. Journal of Personality Assessment, 98(3), 223–237. https://doi.org/10.1080/00223891.2015.1089249
(2020). An updated guideline for assessing discriminant validity. Organizational Research Methods, 15(1), 1–42. https://doi.org/10.1177/1094428120968614
(2011). The temporal course of anxiety sensitivity in outpatients with anxiety and mood disorders: Relationships with behavioral inhibition and depression. Journal of Anxiety Disorders, 25(4), 615–621. https://doi.org/10.1016/j.janxdis.2011.02.001
(2012). lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02
(2014). Psychological inflexibility mediates the effects of self-efficacy and anxiety sensitivity on worry. Spanish Journal of Psychology, 17, Article
E3 . https://doi.org/10.1017/sjp.2014.3(1996). Validation of the Spanish version of the Anxiety Sensitivity Index in a clinical sample. Behaviour Research and Therapy, 34(3), 283–290. https://doi.org/10.1016/0005-7967(95)00074-7
(2007). ASI-3: Nueva escala para la evaluación de la sensibilidad a la ansiedad
[ASI-3: A new scale for the assessment of anxiety sensitivity] . Revista de Psicopatología y Psicología Clínica, 12, 91–104.(2009). Testing structural equation models or detection of misspecifications? Structural Equation Modeling, 16, 561–582. https://doi.org/10.1080/10705510903203433
(2014). Evaluating model fit with ordered categorical data within a measurement invariance framework: A comparison of estimators. Structural Equation Modeling: A Multidisciplinary Journal, 21(2), 167–180. https://doi.org/10.1080/10705511.2014.882658
(1994).
Corrections to test statistics and standard errors in covariance structure analysis . In A. von EyeC. C. CloggEds., Latent variables analysis: Applications for developmental research (pp. 399–419). Sage.(2001). A scaled difference chi-square test statistic for moment structure analysis. Psychometrika, 66, 507–514. https://doi.org/10.1007/BF02296192
(1941). Synthesis of variance. Psychometrika, 6(5), 309–316. https://doi.org/10.1007/bf02288586
(1957). The development of hierarchical factor solutions. Psychometrika, 22, 53–61. https://doi.org/10.1007/BF02289209
(1999). Effects of anxiety sensitivity on anxiety and pain during a cold pressor challenge in patients with panic disorder. Behaviour Research and Therapy, 37(4), 313–323. https://doi.org/10.1016/s0005-7967(98)00139-9
(2002). Structure of the Anxiety Sensitivity Index psychometrics and factor structure in a community sample. Journal of Anxiety Disorders, 1, 33–49. https://doi.org/10.1016/S0887-6185(01)00087-1
(2008). Anxiety sensitivity profile: Predictive and incremental validity. Journal of Anxiety Disorders, 7, 1180–1189. https://doi.org/10.1016/j.janxdis.2007.12.003
(2000). Prospective evaluation of the etiology of anxiety sensitivity: Test of a scar model. Behaviour Research and Therapy, 38(11), 1083–1095. https://doi.org/10.1016/s0005-7967(99)00138-2
(2009). On the use, the misuse, and the very limited usefulness of Cronbach’s alpha. Psychometrika, 74, 107–120. https://doi.org/10.1007/S11336-008-9101-0
(1999). Heritability of anxiety sensitivity: A twin study. American Journal of Psychiatry, 156(2), 246–251. https://doi.org/10.1176/ajp.156.2.246
(2023). Fast and flexible implementations of exploratory factor analysis tools. https://cran.r-project.org/web/packages/EFAtools/EFAtools.pdf
(1997). Gender differences in dimensions of anxiety sensitivity. Journal of Anxiety Disorders, 11(2), 179–200. https://doi.org/10.1016/S0887-6185(97)00005-4
(2014).
Using hierarchical IRT models to create unidimensional measures from multidimensional data . In S. P. ReiseD. A. RevickiEds., Handbook of item response theory modeling: Applications to typical performance assessment (pp. 183–206). Routledge/Taylor & Francis Group.(2018). Using multivariate statistics (7th ed.). Pearson.
(1998a). Anxiety sensitivity: Multiple dimensions and hierarchic structure. Behaviour Research and Therapy, 36, 37–51. https://doi.org/10.1016/S0005-7967(97)00071-5
(1998b). An expanded Anxiety Sensitivity Index: Evidence for a hierarchic structure in a clinical sample. Journal of Anxiety Disorders, 12, 463–483. https://doi.org/10.1016/S0887-6185(98)00028-0
(1992). Conceptualizations of anxiety sensitivity. Psychological Assessment, 4(2), 245–250. https://doi.org/10.1037/1040-3590.4.2.245
(2007). Robust dimensions of anxiety sensitivity: Development and initial validation of the Anxiety Sensitivity Index-3. Psychological Assessment, 19, 176–188. https://doi.org/10.1037/1040-3590.19.2.176
(1989). Anxiety sensitivity: Unitary personality trait or domain-specific appraisals? Journal of Anxiety Disorders, 1, 25–32. https://doi.org/10.1016/0887-6185(89)90026-1
(2004). The greatest lower bound to the reliability of a test and the hypothesis of unidimensionality. Psychometrika, 69(4), 613–625. https://doi.org/10.1007/bf02289858
(2003). Anxiety sensitivity profile: Dimensional structure and relationship with temperament and character. Psychotherapy and Psychosomatics, 72, 217–222. https://doi.org/10.1159/000070786
(1990). Anxiety sensitivity in agoraphobia. Journal of Anxiety Disorders, 4(4), 325–333. https://doi.org/10.1016/0887-6185(90)90029-9
(1998). A retrospective study of the learning history origins of anxiety sensitivity. Behaviour Research and Therapy, 36(5), 505–525. https://doi.org/10.1016/s0005-7967(97)10029-8
(2002). The relation between anxiety sensitivity and attachment style in adolescence and early adulthood. Journal of Psychopathology and Behavioral Assessment, 24(3), 159–168. https://doi.org/10.1023/a:1016058600416
(2012). Dimensions of anxiety sensitivity in the anxiety disorders: Evaluation of the ASI-3. Journal of Anxiety Disorders, 26(3), 401–408. https://doi.org/10.1016/j.janxdis.2012.01.002
(2012). Longitudinal genetic analysis of anxiety sensitivity. Developmental Psychology, 48(1), 204–212. https://doi.org/10.1037/a0024996
(1997). Hierarchical structure and general factor saturation of the Anxiety Sensitivity Index: Evidence and implications. Psychological Assessment, 9(3), 277–284. https://doi.org/10.1037/1040-3590.9.3.277
(2006). Estimating generalizability to a latent variable common to all of a scale’s indicators: A comparison of estimators for ωh. Applied Psychological Measurement, 30(2), 121–144. https://doi.org/10.1177/0146621605278814
(2003). Anxiety sensitivity in six countries. Behaviour Research and Therapy, 7, 841–859. https://doi.org/10.1016/S0005-7967(02)00187-0
(2001). Evaluating differential predictions of emotional reactivity during repeated 20% carbon dioxide-enriched air challenge. Cognition & Emotion, 15(6), 767–786. https://doi.org/10.1080/02699930143000284
