Abstract
Anxiety sensitivity (AS) refers to a person’s tendency to fear anxiety-related symptoms due to the belief that these symptoms may have harmful consequences. The most widely used operationalization of AS in adults is the Anxiety Sensitivity Index (ASI). The factor structure, gender stability, and psychometric properties of the ASI in a sample of Croatian adults (N = 984) were evaluated. Results confirm the multidimensional and hierarchical structure of the ASI, which consisted of three lower-order factors (Physical Concerns, Psychological Concerns, and Social Concerns) and a single higher-order factor, AS. Furthermore, the achievement of normative scores for the ASI in a Croatian adult sample demonstrates the cross-national stability of the ASI. Reliability coefficients for the ASI, Physical Concerns, and Psychological Concerns are high and satisfactory in the total sample and for both genders. Overall, the results confirmed the cross-national stability, gender stability, reliability, and validity of the ASI in a sample of Croatian adults.
Keywords
In clinical psychology, numerous psychological instruments have been translated into different languages and modified for use with specific cultural groups. The majority of such instruments were originally developed in Anglo-Saxon countries and written in English. For researchers in non-English speaking countries, the translation of well-established, standardized instruments is an efficient solution for the lack of available instruments. Using preestablished measures further allows for the cross-cultural comparison of findings as well as use of international comparative research (van Widenfelt, Treffers, de Beurs, Siebelink, & Koudijs, 2005). On the other hand, among professionals, it is generally accepted that knowledge must be derived from research with multiple cultural groups to fully understand risk and protective factors for psychopathology (e.g., American Psychological Association, 2001). Such professional activities, echoing the recommendations of early leaders in this area (Good & Kleinman, 1985), highlight that major gaps in knowledge currently exist as it pertains to understanding the etiology, nature, and prevention/treatment of anxiety psychopathology (Friedman, 1997; McNeil, Porter, Zvolensky, Chaney, & Kee, 2000; Zvolensky, McNeil, Porter, & Stewart, 2001).
Many cultural groups have been disproportionately exposed to various types of social persecution (e.g., McNeil et al., 2000). Such negative life events can promote clinically significant psychological distress and human suffering within and across generations (Clark, Anderson, Clark, & Williams, 1999). Croatia is an Eastern European country currently in a posttransitional period in which society has evolved from a traditional socialist to a liberal capitalist model. Although Croatia is at the end of its transition, the conclusion of these important multilevel changes was postponed by war and the later impact of this war on Croatian society. From 1991 to 1995, apart from one small region of Croatia (Istria and the islands), the entire country was in wartime conditions of varying intensity. The regions of Croatia not affected by war directly became a shelter for refugees. As a result, wartime conditions, postwar consequences, as well as transitional periods were experienced in all regions of Croatia and thus became a source of stress for all citizens. Recent research in a community sample who were directly exposed to war in Croatia and other countries of ex-Yugoslavia (Priebe et al., 2010) documented higher prevalence rates of panic disorder (6% of respondents), posttraumatic stress disorder (18%), and generalized anxiety disorder (10%) than in Western countries (Lieb, Becker, & Altamura, 2005; for review, see Michael, Zetsche, & Jürgen Margraf, 2007). The heightened incidence of anxiety disorders in Croatia demands research attention directed at understanding vulnerability factors. Based on the existing research, anxiety sensitivity (AS) is a vulnerability factor for anxiety psychopathology, especially for panic disorder and posttraumatic stress disorder (for review, see Olutunji & Wolitzky-Taylor, 2009).
AS refers to a person’s tendency to fear anxiety-related symptoms (e.g., racing heart, blushing, having problems breathing, feeling dizzy, shaking) due to the belief that there will be some harmful physical, social, or mental consequences as a result of having these symptoms (Reiss & McNally, 1985). AS is reactivity to anxiety symptoms, and anxiety symptoms are triggered by anxiety. AS has been found to predict panic or anxious responses to challenge and stress beyond trait anxiety (Joiner et al., 2002).
Although AS is a dimensional trait stable over time, it is also state dependent and has been demonstrated to decrease during psychological treatment (Smits, Berry, Tart, & Powers, 2008). Current research indicates that AS refers to individually distinct personality trait distinctive from trait anxiety (McWilliams & Cox, 2001) and that it is related to a variety of anxiety disorders (Ball, Otto, Pollack, Uccello, & Rosenbaum, 2005; Cisler, Reardon, Williams, & Lohr, 2007; Rector, Szacun-Shimizu, & Leybman, 2007; Schmidt, Keough, Timpano, & Richey, 2008; Schmidt, Lerew, & Jackson, 1997) and nonanxiety psychopathology (depression, hypochondria, addiction to nicotine, alcohol, and drugs and chronic pain disorders; Asmundson, 1999; Otto, Pollack, Fava, Uccello, & Rosenbaum, 1995; Stewart, Samoluk, & MacDonald, 1999; Watt & Stewart, 2000).
More recently, the scope of research interests has expanded to better understanding the etiology of AS. Empirical (Taylor, Jang, Stewart, & Stein, 2008) and retrospective (Scher & Stein, 2003; Stewart et al., 2001) studies confirm the importance of environmental factors in the development of AS, although evidence for the importance of genetic factors has also been established (Stein, Jang, & Livesley, 1999; Taylor et al., 2008). Because environmental factors differ across various societies, it would be of a great importance to determine whether manifestations of AS change across nations and cultures.
To accurately understand psychological functioning in any group, it is necessary to employ valid and reliable assessment methodologies (Tanaka-Matsumi, Seiden, & Lam, 1996). The purpose of the present study was to provide an initial psychometric evaluation of a measure of AS in a Croatian sample. To date, there has been no research examining AS in Croatia because of the lack of available instruments.
The most widely used and accepted operationalization of the AS construct in adults is the 16-item version of the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986). In an attempt to improve this instrument, three other versions of the ASI have been developed: Anxiety Sensitivity Index–Revised (Taylor & Cox, 1998a), Anxiety Sensitivity Profile (Taylor & Cox, 1998b), and Anxiety Sensitivity Index–3 (Taylor et al., 2007). However, although there has been some improvement in psychometric properties, there are currently very few studies using these more recent measures.
Research evaluating the psychometric properties of the ASI has demonstrated good internal consistency (α range from .82 to .93) and test–retest reliability over a period of 3 years (.71; Maller & Reiss, 1992). Correlations between the ASI and the State–Trait Anxiety Inventory (STAI; Spielberger, 2000) were between .50 and .60 (Isyanov & Calamari, 2004; McWilliams & Cox, 2001; Smári, Erlendsdóttir, Björgvinsdóttir, & Ágústsdóttir, 2003), whereas correlations between the ASI and the Beck Depression Inventory (BDI-II; Beck, Steer, & Brown, 1996) were somewhat lower (0.41; Smári et al., 2003). Together, these results provide support for the overall validity of the scale.
The factor structure of the ASI has been investigated extensively and has long been at the centre of debate (Deacon & Abramowitz, 2006; Deacon, Abramowitz, Woods, & Tolin, 2003; Hinton, Pich, Safren, Pollack, & McNally, 2005; Zinbarg, Barlow, & Brown, 1997; Zinbarg, Brown, Barlow, & Rapee, 2001; Zinbarg, Mohlman, & Hong, 1999). Even today, the factor structure of the ASI is inconsistent across studies and still remains somewhat controversial. Although AS has been defined as a unitary construct, recent studies have provided support for a hierarchically organized, multidimensional structure that consists of a single higher-order factor (i.e., AS) and a certain number of lower-order factors (Rodriguez, Bruce, Pagano, Spencer, & Keller, 2004). Although a growing body of research has provided supporting evidence for such a multifactorial hierarchical solution, there is still some confusion regarding the exact number of lower-order factors. Several studies have found support for a two-factor solution, describing the factors as Fear of Mental Catastrophe and Fear of Cardiopulmonary Sensation (Schmidt & Joiner, 2002) or, in an alternative interpretation, as Anxiety Sensitivity and Social Concerns (Cintrón, Carter, Suchday, Sbrocco, & Gray, 2005). Other studies have found support for a quadruple lower-order factor solution (Sandin, Chorot, & McNally, 1996; Telch, Shermis, & Lucas, 1989). The majority of studies report three-factor solutions loaded on a single higher-order factor. These factors describe Physical Concerns, Psychological Concerns, and Social Concerns (Rodrigez et al., 2004; Zvolensky et al., 2001). The Physical Concerns factor refers to the fear of physical symptoms related to anxiety (e.g., “It scares me when my heart beats rapidly”). Previous research has shown that this factor has the strongest association with anxiety symptoms and panic disorder but not with depressive mood (Zinbarg et al., 1997). The Psychological Concerns factor refers to concerns related to mental manifestations of anxiety (e.g., “When I am nervous, I worry that I might be mentally ill”). This aspect is less specific for panic disorders (Zinbarg et al., 1997) and has been argued to be a relatively nonspecific measure of agitation (Deacon & Abramowitz, 2006). The Social Concerns factor relates to the fear of publicly observable anxiety symptoms (e.g., “It is important to me not to appear nervous”). This aspect has been the most debated due to its poor stability (Mohlman & Zinbarg, 2000). In many ASI studies, this factor demonstrated a relatively low internal consistency in comparison with other factors and explained the lowest percentage of the total variance of AS (Zvolensky et al., 2001). One possible explanation for this result is that this factor consists of only two items. In light of the generally observed differences in AS among men and women, some studies have additionally examined the stability of the ASI factor structure between men and women (Stewart, Taylor, & Baker, 1997; van Widenfelt, Siebelink, Goedhart, & Treffers, 2002). Although these studies did demonstrate factor stability across genders, further research is still needed to determine whether the factor structures of the ASI are congruent between subsamples of women and men.
Most recently, there is a growing body of research aiming to empirically evaluate the latent structure of AS via taxometrics. Whereas some taxometric studies of AS have reported evidence of a categorical latent structure (Bernstein, Zvolensky, Kotov, et al., 2006; Bernstein, Zvolensky, Stewart, & Comeau, 2007; Bernstein, Zvolensky, Stewart, Comeau, & Leen-Feldner, 2006; Bernstein, Zvolensky, Weems, Stickle, & Leen-Feldner, 2005), others have found evidence of a latent dimension (Broman-Fulks et al., 2008; Broman-Fulks et al., 2010).
In consideration of the multiple inconsistencies in the ASI factor literature, several methodological issues should be noted. First, inadequate sample sizes (less than 200 subjects) might have contributed to various factor analysis results. Blais et al. (2001) reported seven studies of ASI factor structure employing less than 200 subjects. These authors further demonstrated methodological disparities when using different factor analysis techniques (principal component analysis and parallel analysis) in investigating the factor structure of the ASI, a factor that might have influenced the variance in findings across studies.
Second, several studies included samples of different cultural backgrounds (e.g., Puerto Rican, American Indian, Alaska Native, Mexican, Dutch, Spanish, French, and Canadian). Whereas some of these studies were carried out in English speaking countries, others first needed to translate the ASI into the mother tongue of the country where the study was carried out (e.g., Spanish, Farsi, German, French, Hebrew, Greek, Mandarin, Icelandic, and Dutch). Arguably, some of the ASI factor structure differences arising from these studies could be the result of differing cultural characteristics or even of an inadequate translation of the ASI. Although Zvolensky et al. (2003), using the Anxiety Sensitivity Index–Revised, demonstrated that the underlying structure of the AS construct was generally similar across countries, it must be taken into account that only Western European cultures were included in this study. Although the body of research on the ASI in different cultures and nations is growing, there is still much to be understood regarding the use of the ASI in non-Western cultures. To date, there exists no research examining the ASI factor structure and its psychometric properties in any Eastern European country.
Toward this end, the aim of the present study was to provide an initial psychometric evaluation of the ASI in a large, diverse nonclinical sample of Croatian adults. The selection of the ASI was based on its wide recognizability and broad empirical foundation, which, in turn, facilitates the comparison of findings with those from existing studies. The factor structure of the ASI was examined using exploratory and confirmatory analysis. The present study makes a contribution to understanding of the pathogenic influence of potentially stressful events, such as war and transition, to which all Croatian citizens were exposed over the past 20 years. In addition, it may further improve our understanding of the role of environmental factors in the etiology of AS.
Materials and Method
Participants
The study sample comprised a total of 984 adults aged between 18 and 68 years (mean age = 39.2; SD = 11.3), wherein 59% of the sample was female (404 males and 580 females). All subjects were born and continue to live in Croatia.
Procedure
The study was carried out in three Croatian cities in 2008. Subjects were recruited using a convenience sample, in which psychology students participating in the study as investigators were asked to recruit individuals older than 18 years of age from their social setting. Criteria for inclusion in the study were the ability to complete questionnaires independently and no history of psychiatric treatment. All subjects provided informed consent before entering the study.
Instruments
The ASI (Reiss et al., 1986) is a self-report measure of fear of different anxiety symptoms in subjects aged 18 years and older. On a 4-point scale (where 0 is very little and 4 is very much), subjects indicate the degree of aversion to different anxiety symptoms described in 16 items (e.g., “When I notice my heart is beating rapidly, I worry that I might have a heart attack”). The total score is the sum of scores on individual items and varies between 0 and 64. Participants completed the Croatian version of the ASI, which was translated from English by the authors and subsequently back-translated by professional translators.
The BDI-II (Beck et al., 1996) is the most commonly used questionnaire for measuring depression in clinical and research studies. The BDI-II is a 21-item self-report scale measuring the severity of depressive symptoms. Each item comprises four statements reflecting varying degrees of intensity of depressive symptoms. Subjects are asked to choose the statement that best describes the way they have felt over the past 2 weeks. In our investigation, exceptionally high scale reliability was achieved (α = .95).
The STAI (Spielberg, 2000) was developed for the assessment of two related, but logically different constructs: the traits and states of anxiety. Because the present investigation was primarily interested in more stable differences in anxiety and not in participants’ estimation of the degree to which they felt anxious at the time of completing the questionnaire, only the scale of trait anxiety was used. This scale consists of 20 statements in which respondents evaluate their general level of anxiety on a 4-point scale (where 1 = never and 4 = always). Once again, high levels of reliability were reached in our analysis (α = .96).
Results
Normative Data
The mean and the standard deviations of the ASI, STAI-T, and BDI-II achieved in the present study are shown in Table 1. Normative scores for the ASI are M = 19.5 (SD = 10.1) and are similar to those reported for most European American samples (Peterson & Plehn, 1999) and nonclinical Spanish samples (M = 18.8-22.1; Sandin, Chorot, & McNally, 2001), considerably lower than those from a Puerto Rican sample (M = 24.6, SD = 13.9; Cintrón et al., 2005) and higher than those for Alaska Native (M = 17.6, SD = 9.4) and American Indian samples (M = 15.8, SD = 8.4; Zvolensky et al., 2001). In the present sample, scores on the ASI approximate a normal distribution, with the majority of scores (75%) falling between 15 and 35.
Means and Standard Deviations for the Total Sample and Male and Female Subsamples
Note. ASI = Anxiety Sensitivity Index; STAI-T = State and Trait Anxiety Index–Trait; BDI-II = Beck’s Depression Inventory.
p < .001.
A series of t tests between male and female subsamples reveals significant gender differences, where females exhibited significantly higher ASI, STAI-T, and BDI-II scores than males. This result is also shown in Table 1.
Exploratory Factor Analysis
For the purposes of factor analysis, the total sample (n = 984) was randomly split into two separate samples, each consisting of 492 subjects. To determine the factor structure of the ASI, exploratory factor analysis was performed on one of these two samples and separately with female and male subsamples (Table 2). This exploratory analysis consisted of three separate analyses of the main components of all 16 ASI items. Because previous studies suggested a correlation between ASI dimensions (e.g., Asmundson, Frombach, & Hadjistavropoulos, 1998; Cintrón et al., 2005; Rodriguez et al., 2004; Stewart et al., 1997; Zvolensky et al., 2001), a principal component analysis (PCA) with promax (kappa 4) rotation was performed. To determine the number of factors more reliably, we used the following criteria: (a) a Kaiser’s eigenvalue greater than one, (b) Cattell’s scree test, and (c) the results of parallel analysis. Parallel analysis (PA) is a statistical procedure for factor extraction (Hayton, Allen, & Scarpello, 2004) that has higher accuracy than the more commonly used eigenvalue or scree test (Hayton et al., 2004).
Principal Component Analysis of the Anxiety Sensitivity Index Items: Promax-Rotated Factor Loadings (Pattern Matrix) for a Lower-Order Three-Factor Solution (Total Sample and Male and Female Subsamples)
Note. Salient loadings (≥|.35|) are highlighted in boldface. Cronbach’s a coefficients were calculated using unweighted sums of the items that loaded on the factors. Complex items were included only on the factor for which it showed the highest salient loading when calculating coefficients. Eigenvalues for the total sample are as follows: 6.2, 1.3, 1.2, 0.9, 0.9, 0.8, 0.7, 0.6, 0.6, 0.5, 0.5, 0.4, 0.4, 0.4, 0.3, 0.3. Eigenvalues for females are as follows: 6.4, 1.4, 1.2, 0.9, 0.9, 0.8, 0.6, 0.6, 0.6, 0.5, 0.4, 0.4, 0.4, 0.3, 0.3, 0.3. Eigenvalues for males are as follows: 5.8, 1.3, 1.1, 1.0, 0.9, 0.9, 0.9, 0.7, 0.6, 0.5, 0.5, 0.4, 0.4, 0.3, 0.3, 0.2.
On the whole, the results of PCA and PA clearly demonstrate that there are three lower-order factors for the total sample and for each gender separately. The same factor structure was obtained for both genders and for the total sample.
As indicated in Table 2, Factor 1, referring to Physical Concerns, included Items 4, 6, 7, 8, 9, 10, 11, and 14 (eigenvalue = 6.2, 38.62%). Factor 2, referring to Psychological Concerns and an inability to control anxiety symptoms, included Items 2, 3, 12, 13, 15, and 16 (eigenvalue = 1.3, 7.98%). Factor 3 referred to Social Concerns and included Items 1 and 5 only (eigenvalue = 1.2, 7.34%). Only Item 14 (“Unusual body sensations scare me”) showed significant loadings on more than two factors (1 and 2), a finding that points to the complexity of this statement. Indeed, this result might be explained by the somewhat unclear content of this item, where the statement refers to physical sensations that are perceived as unusual but might also include psychological elaboration. This lack of clarity leaves responding to this item open to subjective interpretation. There were no items without any significant loadings on at least one of the factors.
Correlations between factors were as follows: Factor 1 and Factor 2, 0.62; Factor 1 and Factor 3, 0.32; Factor 2 and Factor 3, 0.33. To verify the agreement of corresponding factors between females and males, congruence coefficients were calculated (Kab ranged from .995 to .998; Fulgosi, 1979).
Higher-Order Factor
To determine the higher-order factor, an analysis of the main components was performed using factor scores from the three-factor solution for the total sample and for female and male subsamples. The results of this analysis indicated a single higher-order factor named Anxiety Sensitivity. Table 3 reveals factor loadings, an eigenvalue of 1.8, and an explained variance of 59.9%. To validate the agreement of a single higher-order factor between female and male subsamples, the congruence coefficient was calculated and demonstrates a high level of agreement in the higher-order factor structure between the two subsamples (Kab = .99; Fulgosi, 1979).
Principal Component Analysis of the Higher-Order Structure of the Anxiety Sensitivity Index: Factor Loadings (Component Matrix) for Single-Factor Higher-Order Solutions (Total Sample and Male and Female Subsamples)
Note. Salient loadings (≥|.35|) are highlighted in boldface. Eigenvalues for the total sample are as follows: 1.9, 0.8, 0.4. Eigenvalues for females are as follows: 1.9, 0.8, 0.4. Eigenvalues for males are as follows: 1.9, 0.8, 0.4.
Confirmatory Factor Analysis
On the whole, the PCA and PA conducted in the present study clearly demonstrate a three-factor structure with one higher-order factor, a finding that supports the multidimensionality of the ASI. However, a higher eigenvalue and larger amounts of explained variance for Factor 1, as compared with Factors 2 and 3, raises questions concerning the unidimensionality of the ASI. For this reason, the second split sample (N = 492) was used to examine the fit of a three-factor model with one higher-order factor in comparison with a single-factor model. These results, provided in Table 4, demonstrate that the three-factor model with one higher-order factor is more parsimonious and achieved a better fit index than the one-factor solution.
Fit Indexes for a Triple Higher-Order Model and a Unidimensional Model
Finally, a Schmid–Leiman solution was performed to gain further information and support for a three-factor model as an adequate factorial structure (Wolff & Preising, 2005). Although the results of this analysis, illustrated in Table 5, provide substantial support for a unidimensional structure, there is a stronger support for the three-factor solution. Here, the second-order factor in the hierarchical structure accounts for 68% of the total explained variance. Similarly, the three first-order factors make significant contributions, with each factor accounting for 12%, 10%, and 11% of the total variance explained.
Estimated Parameters of a Hierarchical Model of Three Lower Order Factors Using SL Transformation
Reliability
Internal consistency for a three-factor model was calculated using Chronbach’s alpha coefficients. For the ASI scale in its entirety, the alpha value is .88 (female sample, .87; male sample, 0.89). For Factors 1, 2, and 3, the alpha values are .86 (female, male: .87, .86), .80 (female, male: .82, .78), and 0.45 (female, male: .53, .36), respectively. These coefficients are similar to those obtained in previous research (Cintrón et al., 2005; Reiss, Silverman, & Weems, 2001; Zvolensky et al., 2001). As in these studies, the low alpha value for Factor 3 in the present study can be attributed to the fact that this factor contained only two items.
Correlations With Other Measures
To test convergent validity, partial correlations between ASI, its factors, and STAI-T and BDI-II were calculated for the whole sample. However, in consideration of the inappropriately low reliability of Factor 3, this factor was excluded from this analysis. The decision to calculate partial correlations between measures was in light of the high correlations between STAI-T and BDI-II (r = .68). As is illustrated in Table 6, when controlling for the BDI-II, the ASI and all its factors have a significant positive, but relatively modest correlation with the STAI-T, with the highest correlation achieved between this measure and Factor 2 (Psychological Concerns). When controlling for the STAI-T, the ASI, Factor 1 (Physical Concerns), and Factor 2 have a significant positive but very modest correlation with BDI-II. As was the case for the partial correlations with the STAI-T, the highest correlation is that between the BDI-II and Factor 2.
Partial correlations between the ASI, ASI factors, STAI-T and BDI-II
Note. ASI = Anxiety Sensitivity Index; STAI-T = State and Trait Anxiety Index–Trait; BDI-II = Beck’s Depression Inventory.
p < .001.
Discussion
Using a large nonclinical sample of Croatian adults, the present study evaluated the factor structure, gender stability, and validity of a Croatian version of the ASI. By additionally presenting normative data, this study also contributes to current knowledge about the cross-national stability of the ASI.
Although past investigations have failed to reach a consensus on the issue of normative ASI data, findings have shown that, in the general population, means vary from 14.2 to 22.5, with an overall mean of approximately 19.0 (Peterson & Plehn, 1999). The normative score for the ASI in a Croatian adult sample is M = 19.5 (SD = 10.1), a value similar to those reported for most European American samples (Peterson & Plehn, 1999). Since, to our knowledge, this is the first published study to be conducted in an Eastern European country, it represents a significant contribution to demonstrating the cross-national stability of the ASI. Indeed, given that Croatia is an Eastern European country in a posttransitional period with recent experience in war and postwar consequences, the results of the present study demonstrate that these types of stressful circumstances do not contribute to differing levels of AS at the population level. As such, with regard to potential environmental factors contributing to the development of AS, it might be argued that specific family factors rather than more general societal issues are most influential in the development of higher levels of AS. However, an alternative explanation for these results might be found in a consideration of the structure of the sample in the present study. It has been previously established that the critical period for developing AS is during childhood and early adolescence (Taylor et al., 2008). In the present study, the majority of participants had already reached adulthood during periods of war and transition and, as such, would not have experienced the effects of these circumstances during this critical period. This hypothesis should be tested in future research by examining AS with a Croatian sample of young adults (aged between 25 and 34 years) who would have been between the ages of 5 and 14 years during the war period (1991-1995). In addition, future research should aim to further understand the relationship between specific experiences of war (e.g., direct combat, prisoner of war, war refugee) and AS. Finally, research should aim to further understand the finding, from previous research, demonstrating higher scores on the ASI among individuals with PTSD (Hinton et al., 2005).
A hierarchical, multilevel factor structure, thus far demonstrated predominantly in studies using both English and other language versions of the questionnaire (Norton, De Coteau, Hope, & Anderson, 2004; Zinbarg et al., 1997), was replicated here with a Croatian sample of adults. Exploratory and confirmatory analyses, along with a Schmid–Leiman solution, confirm the multidimensional and hierarchical structure of the ASI consisting of three lower-order factors (Physical Concerns, Psychological Concerns, and Social Concerns) and a single higher-order factor, AS. Thus, the three-factor solution appears to be a good representation of the ASI in the Croatian sample in terms of its parsimony, interpretability, and consistency with other studies (Norton et al., 2004; Zinbarg et al., 1997).
Analysis of the psychometric characteristics of the Croatian versions of the ASI further demonstrates that the instrument is a valid and reliable measure of AS. In addition, it provides a psychometric argument justifying the comparison of the individual expression of AS dimensions in different gender subsamples.
The reliability coefficients for the ASI and its dimensions achieved with the Croatian sample are comparable with values reported in previous studies (Cintrón et al., 2005; Reiss et al., 2001; Zvolensky et al., 2001). Cronbach’s alpha for the ASI (.87-.89), Physical Concerns (.86-.87), and Psychological Concerns (.78-.82) are high and satisfactory in the total sample and for both genders. Also consistent with previous studies (Norton et al., 2004; Stewart et al., 1997; Zinbarg et al., 1997; Zvolensky et al., 2001) are the weak psychometric characteristics of the Social Concerns dimension (α = .36-.53). This was previously argued to be partly a result of the low number of items (two) addressing this factor. As such, although results from this and previous studies achieve good psychometric properties overall, the interpretation of the third dimension of the ASI, Social Concerns, should be carried out very tentatively.
In a comparison with measures of depression symptoms and trait anxiety, the ASI correlates significantly with trait anxiety and depression symptoms and, as such, demonstrates a respectable level of convergent validity with the present sample. All demonstrated correlations are positive, suggesting that high levels of AS are associated with high levels of trait anxiety and a greater severity of recently experienced depression symptoms. A modest partial correlation between trait anxiety and AS confirms that AS is different from trait anxiety and, in the present study, represents a specific vulnerability factor for depression symptoms. In particular, the results demonstrate that ASI shares the greatest portion of the explained variance with trait anxiety and depression symptoms through its Psychological Concerns dimension. This finding represents an additional contribution to the differentiation between AS and trait anxiety.
Although it has been some time since the factor structure and psychometric properties of the ASI have been actively investigated, these issues are still current and relevant topics in research concerning AS. As such, the present study offers a significant contribution to the ongoing debate surrounding these issues. Arguably, still further research and perhaps meta-analytic studies are necessary to provide further clarification and understanding concerning the ASI factor structure.
The present study has several limitations. Although the sample size was sufficiently large, the methods of sample recruitment and data collection did not ensure sample representativeness. Thus, the findings from this study should be verified on more representative samples. In addition, because the present study included nonclinical participants only, the applicability of the findings should be verified on a clinical sample and should examine the influence of individual war experiences of varying intensities. Future research should also examine the incremental validity for the ASI in Croatian samples as well as whether the factor structure can be replicated in other Eastern European countries.
Overall, the results confirm the cross-cultural stability, reliability, and validity of the ASI in adults and for both genders. Furthermore, they demonstrate that the ASI is applicable for use within the Croatian context. In consideration of the previously mentioned specificity of Croatian culture and society, in addition to the presented results, future research with Croatian children and adults could be of great importance and interest to understanding the etiology of AS.
Footnotes
Acknowledgements
Beck Depression Inventory®–Second Edition. Copyright © 1996 by Aaron T. Beck with permission of publisher, Harcourt Assessment, Inc. Croatian translation copyright © 2006 by Aaron T. Beck with permission of publisher, Harcourt Assessment, Inc. Reproduced with permission. All rights reserved. “Beck Depression Inventory” and “BDI” are trademarks of Harcourt Assessment, Inc. registered in the United States of America and/or other jurisdictions.
The authors declared no conflicts of interest with respect to the authorship and/or publication of this article.
The authors disclosed receipt of the following financial support for the research and/or authorship of this article:
The present study was performed as part of a larger project: “Anxiety and Depression in a Life-long Perspective,” funded by the Ministry of Education, Science and Sport of the Republic of Croatia.
