A confirmatory factor analysis of the Beck Anxiety Inventory in African American and European American young adults

https://doi.org/10.1016/j.janxdis.2008.12.003Get rights and content

Abstract

The anxiety literature is particularly sparse as it relates to African Americans, and there are few studies to date that have examined the factor structure of anxiety assessment tools within this population. The current study investigated the original two-factor structure of the Beck Anxiety Inventory (BAI) in addition to two extant factor structures of the BAI in a non-clinical sample of African American and European American young adults. One hundred twenty one European American and 100 African American young adults completed the BAI. Results of a confirmatory factor analysis indicated that the previous factor structures of the Beck Anxiety Inventory do not provide the best fit for either the African American or the European American sample. An exploratory factor analysis revealed that an alternative, two-factor model provided the best fit for the sample, particularly for the African American sample. Implications and suggestions for future research are discussed.

Introduction

The Beck Anxiety Inventory (BAI) has consistently been regarded as a strong tool for measuring cognitive and somatic aspects of self-reported anxiety symptomatology in both clinical and non-clinical populations (Beck, Epstein, Brown, & Steer, 1988; Creamer, Foran, & Bell, 1995; Kumar, Steer, & Beck, 1993; Morin et al., 1999; Osman, Kopper, Barrios, Osman, & Wade, 1997). Examination of the factor structure of the BAI has identified at least two factors that are endemic to the nomological network of anxiety, specifically, cognitive and somatic symptoms (Beck et al., 1988, Creamer et al., 1995, Kumar et al., 1993) whereas other studies have found multiple factors related to the construct of anxiety (Morin et al., 1999, Osman et al., 1997). Although the efficacy of the BAI’s general utility in clinical and non-clinical populations is well-established, very little is known about its use in ethnic minority samples (Contreras, Fernandez, Malcarne, Ingram, & Vaccarino, 2004). The anxiety literature is particularly sparse as it relates to African Americans (Chapman, Kertz, Zurlage, & Woodruff-Borden, 2008; Chapman & Steger, 2008; Heurtin-Roberts, Snowden, & Miller, 1997; Horwath, Johnson, & Hornig, 1994; Lewis-Hall, 1994; Neal & Brown, 1994; Neal & Turner, 1991; Smith, Friedman, & Nevid, 1999). There are no known studies to date that have examined the factor structure of the BAI in an African American sample. The current study conducts a confirmatory factor analysis by examining empirically supported factor structures of the BAI in a sample of African American and European American young adults.

The anxiety literature as it pertains to African Americans remains relatively ambiguous with no clear conclusions being delineated. For example, it has been observed that African Americans have lifetime prevalence rates of anxiety disorders equal to or up to three times higher than that of European Americans (Cooley & Boyce, 2004; Heurtin-Roberts et al., 1997; Neal & Turner, 1991) and are noted as endorsing more agoraphobia than European Americans (Neal & Turner, 1991). In contrast, Last and Perrin (1993) were unable to reveal any significant differences between African Americans and European Americans with regard to simple and social phobias, when the groups were controlled for socioeconomic status. Neal and Turner (1991) concluded that African Americans also may not experience panic disorder in the same manner as their European American counterparts. It is important to note that the experience of these disorders is largely influenced by one’s culture (Heurtin-Roberts et al., 1997, Kirmayer, 2001) and there are likely specific characteristics within this context to be taken into account in the assessment of panic disorder among African Americans versus European Americans. It is also known that African Americans report more somatic symptoms surrounding their experiences with anxiety than do European Americans (Heurtin-Roberts et al., 1997; Neal & Turner, 1991), leading researchers to question the utility of anxiety assessments among ethnically diverse research and clinical populations.

In order to enhance our understanding of anxiety disorders with regard to ethnic minority samples, it is imperative to examine current measures that include somatic-related items, such as the Beck Anxiety Inventory (BAI; Beck et al., 1988). It has been postulated that many cases of anxiety disorders within the African American population go undiagnosed, simply because these individuals are more likely to report symptoms of “nerves” to their primary care provider than to a mental health professional (Kirmayer, 2001, Neal and Turner, 1991) suggesting that African Americans may conceptualize anxiety difficulties much differently than their European American counterparts. Understanding underlying conceptual differences provides clinicians and researchers with a stronger avenue for approaching treatment options with their multicultural clients. Of particular importance may be somatization of anxiety symptoms by African Americans, since previously illustrated empirical work has noted that African Americans are more likely to report these symptoms to a primary care provider. Given that research on anxiety disorders includes predominantly European American samples, it is paramount to investigate whether the utilized measures are valid for use within these populations (Neal & Turner, 1991). Moreover, the deficiency of research pertaining to African American samples (Chapman et al., 2008; Chapman & Steger, 2008; Heurtin-Roberts et al., 1997; Horwath et al., 1994; Lewis-Hall, 1994; Neal & Turner, 1991; Smith et al., 1999) undoubtedly warrants additional examination of the construct of anxiety with existing measures that potentially capture anxiety symptoms endorsed by African Americans (Cooley & Boyce, 2004). Given that the BAI has been particularly adept at identifying somatic facets of anxiety, which African Americans have historically endorsed, it is therefore plausible to investigate anxiety symptoms in a sample of African Americans with the BAI.

The BAI is a 21-item self-report questionnaire, used to measure the degree of severity of anxiety symptoms experienced. Each item is rated on a 4-point Likert-type scale with 0 being “not at all” and 3 being “severely, I could barely stand it.” The scores range from 0–63, with higher scores representing increased severity of the anxiety experiences (Beck et al., 1988). The items are tailored to include psychological complaints as well as somatic symptoms.

Developed in 1988, the primary purpose of the BAI was to bridge the gap that other measures left in the distinction between anxiety disorders and depression (Beck et al., 1988). A major problem in assessing the symptomatology experienced by individuals with anxiety disorders or depression is a measure’s construct validity, more specifically the discriminant validity. The BAI was developed in hopes of alleviating the construct problems inherent in other measures of anxiety and related constructs. By measuring both psychological factors endemic to anxiety as well as somatic complaints, the BAI was able to distinguish between anxiety and depressive symptomatology. In addition to these strengths, the BAI’s utility is further bolstered by having excellent convergent validity in the assessment of these anxiety symptoms (Beck et al., 1988), making it one of the most highly regarded measures for the assessment of anxiety.

When Beck et al. (1988) first developed the BAI, they conducted an exploratory factor analysis on the items included and found that a two-facture structure was the best fit. The items loading on factor one represent the somatic experiences associated with anxiety and the items loading on factor two represent subjective symptoms of anxiety and panic (Beck et al., 1988). There is some particular question with the loading of a few items with the original factor structure in that some items appear to load similarly on both the cognitive and somatic factors (Beck et al., 1988; Hewitt & Norton, 1993). For example, Hewitt and Norton (1993) found that items 7 (“shaky”), 8 (“unsteady”), 12 (“hands trembling”), and 13 (“shaky”) loaded moderately on both the somatic and cognitive factors. Later studies were also not able to replicate the two-factor loading. A series of analyses by Osman and associates found that items on the BAI loaded on a four-factor structure when administered to a population of adolescents (Osman et al., 1997, Osman et al., 2002). Only one study to date has examined the factor structure of the BAI among a primarily non-European American sample (Contreras et al., 2004). In this study, the BAI was administered to a group of Latino university students as well as a group of European American students. Results supported the same two factors structure among the Latino students as well as the European American students. The authors also found that Latino students scored higher than their European American counterparts on nine of the items; five of those items being somatic complaints and four being psychological (Contreras et al., 2004). To date, no study has conducted a factor analysis of the BAI among African Americans.

Hoge et al. (2006) most recently reported that Nepali patients were more likely to endorse somatic items from the BAI when compared to American patients. The BAI was also administered to African young adults in South Africa (Pillay, Edwards, & Dhlomo, 2001). This study reported that 17.8% of their participants scored in the “severe” range on the BAI, illuminating the problem that this measure needs to be empirically tested and examined for its utility in the assessment of anxiety among culturally diverse populations.

Beck and colleagues created the currently accepted factor structure of the BAI through examination of clinical anxiety through the use of items of from three existing measures of anxiety. The original sample was conducted with psychiatric outpatients meeting diagnostic criteria for either anxiety and/or depressive disorders according to the DSM-III criteria. The original item pool was comprised of 86 items and was statistically reduced resulting in the current structure of the BAI, which is comprised of 21 items. The final subsample from the original study carried out by Beck et al. (1988) was 160 subjects. Ethnic/racial group membership was not delineated in this study. An exploratory factor analysis using oblique rotation revealed two factors: somatic symptoms and subjective anxiety/panic. These two factors were moderately and positively correlated (r = .56).

There is a series of existing models that has yielded a two-factor structure similar to the original model described by Beck et al. (1988) (Contreras et al., 2004; Hewitt & Norton, 1993; Kumar et al., 1993, Osman et al., 1997; Steer, Rissmiller, & Ranieri, 1993) with minor exceptions of some items (e.g., somatic items) loading more heavily on the other factor (e.g., cognitive; Hewitt & Norton, 1993). Along these lines, the current investigators utilized these findings as additional rationale to examine the original two-factor structure of the BAI.

Osman et al. (2002) examined the factor structure of the BAI in both a psychiatric inpatient adolescent sample (N = 240) and a non-clinical high school sample (N = 167). The inpatient sample was comprised of 84.2% European Americans, 3.8% African Americans, 1.3% Latinos/Hispanics, and 5% of other ethnic groups. The high school sample included 79.6% European Americans, 8.4% African Americans, 3% Asian Americans, 1.8% Latinos/Hispanics, and 7.2% from other ethnic groups. In attempts to replicate the two-factor structures described above using confirmatory factor analysis, Osman et al. (2002) found that the existing factor structure in their samples was inadequate. A subsequent EFA (principal axis factoring) with orthogonal and oblique rotations supported a four-factor solution for both samples which included neurophysiological, subjective, autonomic, and panic factors. However, because of moderate intercorrelations among these factors (ranging from .31 to .55) further analyses of the principal factors revealed the support for a higher order, single factor structure of the BAI which the authors labeled as anxious arousal. The authors note that the BAI appears to be an effective tool for screening anxiety symptoms in both clinical and non-clinical adolescent samples (Osman et al., 2002).

Morin et al. (1999) sought to examine the psychometric properties of the BAI in a non-clinical sample of 281 older adults recruited from both residential and community settings. Exploratory factor analysis (principal axis factoring) with orthogonal (varimax) rotation yielded a six-factor solution of the BAI which included the following factors: somatic (6 items), fear (3 items), autonomic hyperactivity (3 items), panic (4 items), nervousness (3 items), and motor tension (2 items). The authors note that their findings are consistent with the notion that older adult samples report more somatic complaints than younger populations (Morin et al., 1999). Ethnic/racial group membership was not delineated in this study.

Studies to date underscore the importance of further investigating the utility of the BAI and related measures in diverse populations. The current study examines the factor structure of the BAI in a non-clinical sample of African American and European Americans. Further, we conducted a confirmatory factor analysis of the BAI based upon the three previously described existing factor structures of the BAI. First, the original two-factor structure yielded by Beck et al. (1988) was investigated. Second, given that the BAI was originally developed for use with both adolescents and adults (Beck et al., 1988) the more recent four-factor structure yielded by Osman et al. (2002) in a clinical and non-clinical sample of adolescents was also examined. Since the factor structure of the BAI has yet to be established in either a clinical or non-clinical sample of African Americans and given that the mean age of the current sample was 20 years of age, these factors provided additional rationale for examining the factor structure yielded by Osman and colleagues. Third, the six-factor solution yielded by Morin et al. (1999) in a non-clinical sample of older adults was also examined in the current non-clinical sample. As noted above, existing empirical literature, albeit substantially limited, alludes to African Americans experiencing anxiety symptoms predominantly through somatic complaints (Heurtin-Roberts et al., 1997; Horwath et al., 1994; Neal & Turner, 1991; Smith et al., 1999). Given that older adults have also been shown to endorse more somatic concerns than younger individuals, the six-factor structure yielded by Morin et al. (1999) was examined to assess whether somatically focused experience/expression of anxiety would better characterize the factor structure in a diverse sample.

Section snippets

Participants

The current study included a total of 221 undergraduate students from a medium-large public university. Students were recruited from two sources, specifically an introduction to psychology course (N = 130) and an introductory-level Pan African Studies course (N = 91). Both African American and European American participants were derived from these two courses. The sample was comprised of 71 males and 150 females with a mean age of 20 years. Forty-five percent of the sample was African American

Demographic comparisons

African American and European American participants were compared on demographic variables and the BAI measure. Results of the demographic comparisons indicated that African American and European American participants significantly differed with respect to age and income with African Americans reporting lower income (χ2 = 25.21, p < .001) and being older than the European Americans in the current sample (t = −4.03, p < 001). The two groups did not significantly differ on living arrangements (t = 3.35)

Discussion

To date, the current study is the first to examine the factor structure of the BAI in a sample of African American young adults as compared to European American young adults. The primary aim of the current study was to examine the existing factor structure of the BAI (two factors) via CFA in order to determine if the original factor structure holds in a diverse sample of African American and European American young adults. Further, we investigated whether other published factor structures of

References (30)

  • L.K. Chapman et al.

    A confirmatory factor analysis of specific phobia domains in African American and Caucasian American young adults

    Journal of Anxiety Disorders

    (2008)
  • M. Creamer et al.

    The Beck Anxiety Inventory in a non-clinical sample

    Behaviour Research and Therapy

    (1995)
  • A.T. Beck et al.

    Manual for the Beck Anxiety Inventory

    (1990)
  • A.T. Beck et al.

    An inventory for measuring clinical anxiety: psychometric properties

    Journal of Consulting and Clinical Psychology

    (1988)
  • P.M. Bentler

    Comparative fit indexes in structural models

    Quantitative Methods in Psychology Bulletin

    (1990)
  • K.A. Bollen

    A new incremental fit index for general structural equation models

    Sociological Methods and Research

    (1989)
  • L.K. Chapman et al.

    Race and religion: Differential prediction of anxiety symptoms by religious coping in African American and European American young adults

    Depression and Anxiety

    (2008)
  • F. Chen et al.

    An empirical evaluation of the use of fixed cutoff points in RMSEA test statistic in structural equation modeling

    Sociological Methods and Research

    (2008)
  • S. Contreras et al.

    Reliability and validity of the beck depression and anxiety inventories in Caucasian Americans and Latinos

    Hispanic Journal of Behavioral Sciences

    (2004)
  • M.R. Cooley et al.

    An introduction to assessing anxiety in child and adolescent multiethnic populations: challenges and opportunities for enhancing knowledge and practice

    Journal of Clinical Child and Adolescent Psychology

    (2004)
  • S. Heurtin-Roberts et al.

    Expressions of anxiety in African Americans: ethnography and the epidemiological catchment area studies

    Culture, Medicine, and Psychiatry

    (1997)
  • P.L. Hewitt et al.

    The Beck Anxiety Inventory: a psychometric analysis

    Psychological Assessment

    (1993)
  • E. Horwath et al.

    Epidemiology of panic disorder in African-Americans

    The American Journal of Psychiatry

    (1994)
  • E.A. Hoge et al.

    Cross-cultural differences in somatic presentation in patients with generalized anxiety disorder

    The Journal of Nervous and Mental Disease

    (2006)
  • R.H. Hoyle et al.

    Formulating clinical research hypotheses as structural equation models: a conceptual overview

    Journal of Consulting and Clinical Psychology

    (1994)
  • Cited by (48)

    • Mental Health and Treatment Considerations for Black Americans

      2022, Comprehensive Clinical Psychology, Second Edition
    • Anxiety, Depression, and Anger in Bruxism: A Cross-sectional Study among Adult Attendees of a Preventive Center

      2021, Psychiatry Research
      Citation Excerpt :

      Depressive symptoms over the previous 2 weeks were assessed through the Patient Health Questionnaire-9 (PHQ-9). ( Chapman et al., 2009, Kroenke et al., 2001, Santos et al., 2013) This 9-item instrument is scored in four ordinal responses, from 0 ('never') to 3 ('almost every day'), with possible scores ranging from 0 to 27. The cut-off point ≥ 10 suggests a major depressive episode.

    • The impact of childhood family functioning on anxious, depressive, and obsessive-compulsive symptoms in adulthood among African Americans

      2015, Journal of Obsessive-Compulsive and Related Disorders
      Citation Excerpt :

      The measure includes 21 items rated on a scale of 0–3 and has been found to possess high internal consistency, test–retest reliability, and concurrent and discriminant validity (Beck, Epstein, Brown, & Steer, 1988; Hewitt & Norton, 1993). The BAI has also been shown to have sufficient psychometric properties in African Americans (Chapman & Woodruff-Borden, 2009; Chapman, Williams, Mast, & Woodruff-Borden, 2009) and shows strong internal reliability within the current sample (α=0.93). The MEIM (Phinney, 1992) is 20-item questionnaire that assesses ethnic identity across groups and was only measured in the Penn sample.

    View all citing articles on Scopus
    View full text