The Anxiety Sensitivity Index - Revised: psychometric properties and factor structure in two nonclinical samples

https://doi.org/10.1016/S0005-7967(03)00065-2Get rights and content

Abstract

Anxiety sensitivity (AS) is the fear of anxiety-related sensations based on beliefs about their harmful consequences. Despite its status as the most popular measure of AS, the anxiety sensitivity index is too abbreviated to adequately measure the somatic, cognitive, and social facets of the construct. The Anxiety Sensitivity Index - Revised (ASI-R) is a revised and expanded version of the ASI that was developed to improve the assessment of AS and its dimensions. The present study was conducted to examine the psychometric properties and factor structure of the ASI-R. Two large undergraduate samples completed a psychometric assessment package that included the ASI-R and measures of anxiety, depression, and related constructs. Exploratory factor analysis revealed four lower-order ASI-R factors: (1) beliefs about the harmful consequences of somatic sensations; (2) fear of publicly observable anxiety reactions; (3) fear of cognitive dyscontrol; and (4) fear of somatic sensations without explicit consequences. These factors loaded on a single, higher-order factor. Correlations between the ASI-R factors and related variables were consistent with AS theory. Results across both samples in the present study were highly similar. The strengths and limitations of the ASI-R are discussed, and the implications of our findings for the nature and measurement of AS are considered.

Introduction

Anxiety sensitivity (AS) refers to the fear of anxiety-related sensations, which is thought to arise from beliefs that these symptoms have harmful physical, psychological, or social consequences (Reiss & McNally, 1985). AS is considered a dispositional trait that amplifies fear and other anxiety reactions and places individuals at risk for the development of anxiety-related conditions, particularly panic disorder (Reiss, 1991). AS is distinct from trait anxiety (i.e. the tendency to respond with fear to a wide range of stressors) and describes a more specific tendency to fearfully respond to one’s own anxiety symptoms. The construct validity of AS is supported by an impressive body of research (summarized in Taylor, 1999) that documents the role of AS in anxiety and panic. For example, research has reliably demonstrated that AS distinguishes panic disorder from other anxiety disorders (Apfledorf, Shear, Leon, & Portera, 1994), predicts fearful responding to panic symptom provocation procedures (e.g. Rapee, Brown, Antony, & Barlow, 1992), and predicts prospective development of panic attacks (e.g. Schmidt, Lerew, & Jackson, 1997).

The factor structure of AS has important implications for the nature of AS and its role in anxiety-related psychopathology. Factor analysis allows researchers to study the basic mechanisms of AS (Taylor & Cox, 1998a), because distinct factors may correspond to distinct mechanisms (Cattell, 1978). Different AS mechanisms may have distinct causes (e.g. learning experiences) that may lead to specific anxiety reactions. For example, as a result of observing a family member die of a heart attack, an individual might develop a fear of cardiac sensations that could trigger a panic attack when that individual experiences heart palpitations (Cox, 1996). There is converging evidence from recent factor analytic studies that the Anxiety Sensitivity Index (ASI; Reiss, Peterson, Gursky, & McNally, 1986), the most commonly used measure of AS, measures three replicable, lower-order factors: (1) fear of somatic sensations, (2) fear of cognitive dyscontrol, and (3) fear of publicly observable anxiety symptoms (see Zinbarg, Mohlman and Hong, 1999, for a review). These lower-order factors appear to be hierarchically arranged beneath a single higher-order factor (i.e. general AS). Studies examining the correlates of ASI factors have confirmed the importance of a multidimensional perspective of AS. The ASI fear of somatic sensations factor is most strongly associated with a diagnosis of panic disorder (Zinbarg, Barlow and Brown, 1997, Taylor, Koch, Woody and McLean, 1996) and is the strongest predictor of fearful responding to panic symptom provocation procedures (Zinbarg, Brown, Barlow, & Rapee, 2001). The AS fear of cognitive dyscontrol factor appears less specific to panic disorder and more sensitive to depression (Blais, Otto, Zucker, McNally, Schmidt, Fava and Pollack, 2001, Taylor, Koch, Woody and McLean, 1996). The third factor from the ASI, fear of publicly observable anxiety symptoms, appears to be associated with negative evaluation sensitivity and a diagnosis of social phobia (McWilliams, Stewart and MacPherson, 2000, Zinbarg, Barlow and Brown, 1997). Knowledge about AS has been significantly enhanced by studies on the association between AS factors and various types of psychopathology, and important theories about the role of AS dimensions in the development of panic (e.g. Cox, 1996) await empirical validation.

Current conceptualizations of the factor structure of AS (e.g. Zinbarg, Mohlman & Hong, 1999) are based almost exclusively on factor analytic studies of the ASI. The ASI is a 16-item self-report scale that was constructed to measure what was originally conceptualized as a unitary construct (Reiss, Peterson, Gursky & McNally, 1986). Because the ASI contains a relatively small number of items, most of which measure fears of somatic sensations (e.g. Stewart, Taylor, & Baker, 1997), the scale is too abbreviated to adequately measure the major AS factors. For example, the ASI has too few items to ascertain whether the ‘fear of somatic sensations’ factor may actually consist of several factors, such as fears of cardiac symptoms and fears of gastrointestinal symptoms (Taylor & Cox, 1998a). Further, the ‘social concerns’ factor of the ASI reliably consists of only two items (e.g. Deacon & Valentiner, 2001), both of which have questionable face validity for the AS construct (e.g. ‘It’s important for me not to appear nervous’). The wording of several ASI items is also problematic; some items are ambiguous (e.g. ‘Unusual body sensations scare me’), whereas some items seem to assess constructs other than AS (e.g. ‘It is important for me to stay in control of my emotions’). In support of these criticisms, a recent study by Blais et al. (2001) showed that five particularly problematic ASI items (31% of the scale) could be deleted without reducing the scale’s construct validity. Taken together, the numerous limitations of the ASI caution against reliance on this instrument for making theoretical claims about the AS construct. Clearly, researchers interested in measuring AS factors would be well advised to look for alternatives to the ASI.

The Anxiety Sensitivity Index - Revised (ASI-R; Taylor & Cox, 1998b) was developed to more comprehensively measure the lower-order factors of AS. The 36-item ASI-R retains the same instructions and response format as the ASI, and contains 10 of the ASI’s original 16 items. Six items from the original ASI with problematic content were eliminated, including three of the five psychometrically deficient items identified by Blais et al. (2001). Drawing on domains identified in the ASI factor analytic literature, the authors constructed the ASI-R to measure fears of cardiovascular, respiratory, gastrointestinal, publicly observable, dissociative and neurological, and cognitive dyscontrol anxiety symptoms. In a sample of 155 psychiatric outpatients, Taylor & Cox, 1998b) found that the ASI-R measured four lower-order AS factors (in addition to a higher-order, general AS factor): (1) fear of respiratory symptoms, (2) fear of publicly observable anxiety reactions, (3) fear of cardiovascular symptoms, and (4) fear of cognitive dyscontrol. The lower-order ASI-R factors demonstrated theoretically consistent relationships with criterion variables such as measures of anxiety and depression and psychiatric diagnosis. The ASI-R higher-order factor was correlated with the ASI at r=0.94, indicating that both indices measure the same construct. The results of Taylor & Cox, 1998b) suggest that the ASI-R is a promising instrument for measuring AS. It is also possible that given its superior content validity, the ASI-R is better suited than the ASI for use in studies of AS factors.

Despite its appeal as a potentially improved measure of AS and its dimensions, the ASI-R’s psychometric properties have not been adequately evaluated. At the time of this writing, no follow-up study to Taylor & Cox, 1998b) has appeared in the literature. Taylor and Cox’s factor analytic results need to be replicated, particularly given that the four-factor structure they obtained did not correspond to the six factors the scale was designed to measure. Further, the psychometric properties of the ASI-R have not been evaluated in a nonclinical population. Although research on the original ASI suggests that the dimensional structure of AS is invariant across clinical, nonclinical populations, and community samples (Zinbarg, Mohlman and Hong, 1999, Schmidt and Joiner, 2002), the extent to which this is holds true for the ASI-R is unknown. Given the potential for research on nonclinical samples to provide key insights about the role of AS in the development of anxiety and panic (e.g. Schmidt, Lerew & Trakowski, 1997), the establishment of an improved measure of AS would be an important development for research efforts in this area. To address these concerns, the present study examined the psychometric properties and construct validity of the ASI-R in a nonclinical, undergraduate sample. We hypothesized that, consistent with the results of Taylor & Cox, 1998b), the ASI-R would consist of four replicable lower-order factors assessing fears of respiratory, cardiovascular, publicly observable, and cognitive dyscontrol anxiety symptoms. We further predicted that these lower-order factors would load on to a single higher-order factor, thus supporting the hierarchical structure of AS. Finally, we hypothesized that the ASI-R and its lower-order factors would demonstrate a pattern of theoretically consistent relationships with related variables (e.g. agoraphobic cognitions, fear of negative evaluation).

Section snippets

Participants

The sample consisted of 558 college students recruited from introductory psychology courses at University of North Carolina at Chapel Hill. The sample was 75.1% female with a mean age of 19.0. Four hundred and twenty participants (75.3%) identified themselves as White/Caucasian, followed by 82 Black/African Americans (14.7%), 32 Asians or Pacific Islanders (5.7%), and 24 participants (5.2%) of other, multiple, or unreported ethnicities.

Anxiety Sensitivity Index - Revised (ASI-R)

The ASI-R (Taylor & Cox, 1998b) is a 36-item, expanded

Reliability and item-level analyses

The mean ASI-R total score was 25.7 (S.D.=19.6). ASI-R total scores for women (M=26.7, S.D.=19.5) were significantly higher than those for men (M=22.1, S.D.=19.4), t(556)=−2.51, P<0.05. Given that the scale consisted of 36 items, these mean ASI-R total scores indicate that participants tended to indicate either ‘very little’ or ‘a little’ agreement with the scale items. Means and standard deviations for the ASI-R items are presented in Table 1. Mean scores on 25 out of 36 items were below 1.0

Discussion

Findings from study 1 generally replicated those reported by Taylor & Cox (1998b). The ASI-R was composed of lower-order factors assessing fears of somatic, publicly observable, and cognitive dyscontrol anxiety symptoms. These factors were statistically reliable and demonstrated theoretically consistent relationships with related variables. The two ASI-R somatic factors, however, were less replicable than the other factors and diverged from the results reported by Taylor & Cox (1998b) in an

Study 2: Method

The study questionnaires were administered to a second sample of undergraduate students recruited from introductory psychology courses at University of North Carolina at Chapel Hill. This sample consisted of 444 participants, including 332 women (74.8%), with a mean age of 19.0. The sample was 77% White/Caucasian (n=342), followed by 56 Black/African Americans, 21 Asians or Pacific islanders (4.7%), and 25 participants (5.6%) of other, multiple, or unreported ethnicities. The measures and

Reliability and item-level analyses

Mean ASI-R total scores were 25.4 (S.D.=18.7), and were marginally higher for women (M=26.3, S.D.=19.9) than for men (M=22.7, S.D.=14.4), t (442)=1.77, P<0.10. The scale demonstrated excellent internal consistency (α=0.94). Each item had an adequate corrected item-total correlation (M=0.55, range=0.30 to 0.68).

Factor structure of the ASI-R

Exploratory factor analysis was used to examine the ASI-R’s factor structure. Although confirmatory factor analysis (CFA) is sometimes used in similar situations, at least three caveats

Discussion

The present study evaluated the psychometric properties, factor structure, and construct validity of the ASI-R (Taylor & Cox, 1998b) in two samples with a combined total of 1002 participants. Our findings indicate that the ASI-R is highly internally consistent with all items correlating sufficiently with the total scale. These results stand in contrast to research indicating that the original ASI (Reiss, Peterson, Gursky & McNally, 1986) contains numerous items with unacceptable psychometric

Acknowledgements

This research was supported in part by NIMH NRSA grant number 1 F31 MH67334-01 to the third author.

References (44)

  • S Taylor et al.

    Anxiety sensitivity, trait anxiety, and the anxiety disorders

    Journal of Anxiety Disorders

    (1991)
  • W.W.K Zung

    A rating instrument for anxiety disorders

    Psychosomatics

    (1971)
  • W.J Apfledorf et al.

    A brief screen for panic disorder

    Journal of Anxiety Disorders

    (1994)
  • M.A Blais et al.

    The anxiety sensitivity index: Item analysis and suggestions for refinement

    Journal of Personality Assessment

    (2001)
  • R.B Cattell

    The scientific use of factor analysis in the behavioral and life sciences

    (1978)
  • D.L Chambless et al.

    Assessment of fear of fear in agoraphobics: The body sensations questionnaire and the agoraphobic cognitions questionnaire

    Journal of Consulting and Clinical Psychology

    (1984)
  • B.J Deacon et al.

    Dimensions of anxiety sensitivity and their relationship to nonclinical panic

    Journal of Psychopathology and Behavioral Assessment

    (2001)
  • R.L Gorsuch

    Factor analysis

    (1983)
  • E Guadagnoli et al.

    Relation of sample size to the stability of component patterns

    Psychological Bulletin

    (1988)
  • R.O Jegede

    Psychometric attributes of the self-rating anxiety scale

    Psychological Reports

    (1977)
  • S.O Lilienfeld et al.

    Comment on Holloway and McNally’s (1987) ‘Effects of anxiety sensitivity on the response to hyperventilation’

    Journal of Abnormal Psychology

    (1989)
  • Cited by (0)

    View full text