Negative-reinforcement drinking motives mediate the relation between anxiety sensitivity and increased drinking behavior

Personality and Individual Differences 31 (2001) 157±171

www.elsevier.com/locate/paid

Negative-reinforcement drinking motives mediate the relation between anxiety sensitivity and increased drinking behavior Sherry H. Stewart a,*, Michael J. Zvolensky b, Georg H. Eifert c a

Department of Psychology, Dalhousie University, Halifax, Nova Scotia, Canada BBH 4J1 b Department of Psychiatry and Human Behavior, Providence, Rhode Island, USA c Department of Psychology, West Virginia University, Morgantown, West Virginia, USA

Received 18 February 2000; received in revised form 23 October 2000; accepted 6 November 2000

Abstract We examined whether certain ``risky'' drinking motives mediate the previously established relation between elevated anxiety sensitivity (AS) and increased drinking behavior in college student drinkers (n=109 women, 73 men). Speci®cally, we administered the Anxiety Sensitivity Index (ASI), Revised Drinking Motives Questionnaire, and a quantity-frequency measure of typical drinking levels. Participants were parceled according to high (n=30), moderate (n=29), and low (n=34) AS levels. As expected, high AS participants reported a higher typical weekly drinking frequency than the low and moderate AS students regardless of gender. Similarly, high AS participants (particularly high AS men) reported a higher yearly excessive drinking frequency than low AS students. Only the negative reinforcement motives of Coping and Conformity were found to independently mediate the relations between AS and increased drinking behavior in the total sample. High AS women's greater drinking behavior was largely explained by their elevated Coping Motives, while heightened Conformity Motives explained the increased drinking behavior of high AS men. Finally, associations between AS and increased drinking behavior in university students were largely attributable to the ``social concerns'' component of the ASI. We discuss the observed relations with respect to the psychological functions of drinking behavior that may portend the development of alcohol problems in young adult high AS men and women. # 2001 Elsevier Science Ltd. All rights reserved. Keywords: Anxiety sensitivity; Drinking motives; Drinking levels

* Corresponding author. Tel.: +1-902-494-3793; fax: +1-902-494-3417. E-mail address: [email protected] (S.H. Stewart). 0191-8869/01/$ - see front matter # 2001 Elsevier Science Ltd. All rights reserved. PII: S0191-8869(00)00213-0

158

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

1. Introduction Anxiety sensitivity (AS) is an individual di€erence variable involving beliefs that anxiety-related sensations are associated with negative consequences, such as physical illness, mental incapacitation, or social embarrassment (McNally, 1996; Stewart, Taylor, & Baker, 1997). Elevated levels of anxiety sensitivity have been observed among individuals with diagnosed anxiety disorders including patients with panic disorder (e.g. Stewart, Knize, & Pihl, 1992) and social phobia (e.g. Ball, Otto, Pollack, Uccello, & Rosenbaum, 1995). Longitudinal research has demonstrated that high AS serves as a pre-morbid vulnerability factor for the development of panic attacks and anxiety disorders (Maller & Reiss, 1992; Schmidt, Lerew, & Jackson, 1997). In addition to AS's association with anxiety disorders, theorists have argued that it may be a risk factor for greater alcohol use behavior and for alcohol use disorders (McNally, 1996; Stewart, Samoluk, & MacDonald, 1999). Given their extreme fear of anxiety, high AS individuals should be more motivated than others to learn to use any drug (e.g. alcohol) which is capable of reducing anxious emotions or arousal sensations (Stewart et al., 1999). The fact that high AS individuals are particularly sensitive to the fear dampening e€ects of alcohol (e.g. Conrod, Pihl, & Vassileva, 1998; MacDonald, Baker, Stewart, & Skinner, 2000; Stewart & Pihl, 1994) suggests that they should have increased opportunity to learn about such rewarding consequences of drinking. Empirical research suggests high AS levels may be related to alcohol use disorders by virtue of their association with greater drinking levels and/or ``risky'' reasons for drinking (see Stewart et al., 1999, for review). Indeed, AS levels are positively associated with self-reported weekly drinking rates in panic disorder patients (Cox, Swinson, Shulman, Kuch, & Reichman, 1993), and with self-reported weekly drinking rates and yearly excessive drinking episodes in university women (Stewart, Peterson, & Pihl, 1995). Despite this relatively-well established relation between AS and increased drinking behavior, it is not clear whether the higher self-reported weekly drinking levels of high AS individuals are due to a greater drinking quantity, frequency, or both (cf. Stewart, Angelopoulos, Baker, & Boland, 2000; Vogel-Sprott, 1983). Moreover, since the Stewart et al. (1995) study focused exclusively on female university students, it remains unknown whether the relation of AS to increased drinking behavior varies across gender. In addition to personality variables such as AS, drinking motives also appear to be related to levels of alcohol use, and to drinking problems (Cooper, 1994; Cooper, Frone, Russell, & Mudar, 1995; Cooper, Russell, Skinner, & Windle, 1992). For example, Coping, Conformity, Enhancement, and Social Motives all predict unique aspects of drinking behavior (e.g. Carrigan, Samoluk & Stewart, 1998). Speci®cally, higher levels of alcohol use are associated with Coping and Enhancement Motives (internal motives) as compared to Social Motives (external motives; Cooper, 1994; Cooper et al., 1992). Further, Coping and Conformity Motives (negative reinforcement motives) directly predict drinking problems, whereas Enhancement and Social Motives (positive reinforcement motives) do not (Cooper, 1994; Cooper et al., 1992). Given these patterns of relations to drinking behaviors and drinking outcomes, Coping, Conformity, and Enhancement Motives have been described as relatively ``risky'' reasons for drinking (e.g. Cooper, 1994). High AS individuals do display ``risky'' drinking motives as measured by the three-factor Drinking Motives Questionnaire (DMQ; Cooper et al., 1992) or its revised four-factor version

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

159

(DMQ-R; Cooper, 1994). In particular, AS is an independent predictor of Coping and Conformity Motives (Conrod et al., 1998; Stewart, Karp, Pihl, & Peterson, 1997; Stewart & Zeitlin, 1995; Stewart, Zvolensky, & Eifert, 1999). Nonetheless, it is not clear whether drinking motives mediate previously established relations between AS and drinking behavior (cf. Cooper et al., 1995; Stewart & Devine, 2000; Stewart, Loughlin, & Rhyno, 2001). A ``mediator variable'' explains the relationship between a predictor and a given criterion, thereby permitting increased understanding of why the predictor±criterion relation exists (Baron & Kenny, 1986). Researchers have proposed that drinking motives are the ®nal common pathway to alcohol use and abuse, through which other risk factors (such as personality factors) exert their in¯uences on drinking behavior (e.g. Cooper, 1994; Cox & Klinger, 1988). A mediational perspective would suggest that high AS individuals drink more than others at least in part because they drink primarily to relieve aversive states (i.e., for Coping and/or Conformity Motives). Moreover, because AS presents somewhat di€erently in women and men (e.g., men's greatest anxiety-related concerns falling in the social concerns domain; Stewart, Conrod, Gignac, & Pihl, 1998; Stewart, Taylor, & Baker 1997) the relative importance of Conformity versus Coping motives in explaining high AS individuals' greater drinking behavior might vary across gender. A ®nal area requiring further investigation pertains to the speci®c relations of the various lower-order AS components with increased drinking behaviors. It is now fairly well established that AS is organized hierarchically with several lower-order components each of which loads on a higher-order general AS factor (e.g. Zinbarg, Barlow, & Brown, 1997). The most replicable lowerorder AS components appear to be physical concerns, psychological concerns, and social concerns about the consequences of anxiety-related sensations (e.g. McWilliams, Stewart, & MacPherson, 2000; see also review in Zinbarg, Mohlman, & Hong, 1999). Only one study to date has examined relations between these lower-order AS components and drinking behavior. McWilliams and Asmundson (1999) recently showed that social concerns, but not psychological concerns or physical concerns, were predictive of increased drinking behavior among a large sample of university students. The present study had several aims. First, we sought to determine whether AS levels are associated with particular aspects of drinking behavior in a large, mixed-gender, undergraduate sample. We hypothesized that high AS levels would be associated with overall greater drinking behavior compared to moderate and low AS levels. We explored whether this predicted relationship varied across drinking behavior measures (i.e. typical drinking quantity, typical weekly drinking frequency, and yearly excessive drinking episodes) and/or across gender. Second, we examined the role of certain ``risky'' drinking motives in mediating the hypothesized relations between high AS and increased drinking behavior. We hypothesized that Coping and Conformity drinking motives would independently mediate the tendency of high AS individuals to engage in increased drinking behavior. We also examined the mediating roles of Coping and Conformity motives separately in each gender group to determine whether the particular drinking motive mediators might di€er between university men and women. Finally, we examined the associations between each lower-order AS component and the three drinking behavior measures. Of the three previously identi®ed lower-order AS components of physical, psychological, and social concerns (e.g. Stewart, Taylor, & Baker 1997), we expected that AS social concerns would show the strongest relations with increased drinking behavior (cf. McWilliams & Asmundson, 1999).

160

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

2. Method 2.1. Participants A total of 205 undergraduate students from West Virginia and Dalhousie Universities voluntarily participated in this study. Participants were primarily Caucasian. Of these 205 students, 23 (11%; 15 females, 8 males) indicated that they had not consumed any alcohol within the last year, and the remaining 182 (89%; 109 females, 73 males) were self-classi®ed as ``drinkers''. Only the data from these 182 drinkers were included in further analyses, since completion of the DMQ-R (Cooper, 1994) requires that the respondent have consumed at least some alcohol in the last year. Drinkers had a mean age of 22.3 (S.D.=5.4) years. 2.2. Materials 2.2.1. Revised Drinking Motives Questionnaire (DMQ-R) The DMQ-R (Cooper, 1994) is a 20-item self-report measure of the degree to which individuals use alcohol for four conceptually and empirically distinct reasons: Coping Motives (e.g. ``To forget about your problems''), Conformity Motives (e.g. ``To ®t in with a group you like''), Enhancement Motives (e.g. ``Because its fun''), and Social Motives (e.g. ``To celebrate a special occasion with friends''). Respondents indicate their relative frequency of alcohol use for each of the indicated reasons, on a 5-point scale with anchors of 1 (almost never/never) and 5 (almost always/always). The DMQ-R has sound psychometric properties (Cooper). 2.2.2. Anxiety Sensitivity Index (ASI) The ASI (Peterson & Reiss, 1992) is a 16-item self-report questionnaire that assesses an individual's level of fear of anxiety-related symptoms (e.g. rapid heart beat, dizziness, shaking) based on the belief that such sensations have negative consequences. Respondents rate the degree to which they agree or disagree with each item on a 5-point scale with anchors of 0 (very little) and 4 (very much). The ASI has sound psychometric properties (Peterson & Reiss). 2.2.3. Demographics and drinking behavior measure An author-compiled questionnaire was used to assess age, gender, and the current year of students' university program. Frequency and quantity of alcohol consumption were assessed on the same questionnaire in a manner previously described by Stewart et al. (1995). In regard to frequency, students reported the number of occasions per week on which they normally consumed alcohol; those who consumed alcohol on less than one occasion weekly, estimated monthly or yearly frequency. In regard to quantity, students indicated the average number of standard alcoholic beverages (one drink=12 oz. beer, 1 oz. hard liquor, or 4 oz. wine) they normally consumed per drinking occasion. We used methods recommended by Sobell and Sobell (1990) for enhancing the accuracy of these retrospective self-reports (e.g. drinking items were presented in an open-ended fashion, con®dentiality was assured). We calculated an excessive drinking frequency measure, de®ned as the average number of times per year each student drank four or more drinks per drinking occasion (cf. Stewart, Angelopoulos et al., 2000). A minimum of four drinks per occasion was used as the cuto€ given this is the minimal drinking

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

161

quantity associated with adverse outcomes (Wechsler, Davenport, Dowdall, Moeykens, & Castillo, 1994). 2.3. Procedure After providing informed consent, participants completed questionnaires anonymously in the above-listed order. Participants were debriefed as to study objectives and hypotheses prior to their departure. 3. Results 3.1. Extreme groups comparisons In order to compare the present results with previous studies in which high, moderate, and low AS groups have been compared in terms of drinking behavior patterns (e.g. Stewart et al., 1995), we ®rst examined scores on the three drinking behavior indices using extreme AS group comparisons. Participants were selected from among those in the larger sample of 182 drinkers to ®ll three AS groups (low, moderate, and high AS) based on a comparison of their ASI scores with the sample mean (i.e. M=16.3; S.D.=9.1). All of those students who scored at least one S.D. above the sample ASI mean were selected for inclusion in the high AS group (i.e. ASI score 5 25; n=30). All of those students who scored at least one S.D. below the sample ASI mean were selected for inclusion in the low AS group (i.e. ASI score4 7; n=34). A group of equivalent size was created to ®ll the moderate AS group by selecting students who closely approximated the sample ASI mean (i.e. range: 15±18 inclusive; n=29). Since the three AS groups failed to di€er signi®cantly in gender composition [w2 (2)=3.22, n.s.; Table 1], the three AS groups were further divided by gender to create a total of six cells (high AS men, high AS women, moderate AS men, moderate AS women, low AS men, low AS women). A series of 32 (AS group  gender) analyses of variance (ANOVAs) on age and current year of university program failed to reveal any signi®cant e€ects (all P's > 0.05). A 32 ANOVA on total ASI scores revealed a signi®cant main e€ect of gender [F (1, 87)=22.30, P< 0.001] with women scoring higher overall than men [Ms (and S.D.s)=18.8 (11.7) vs. 14.9 (11.3), respectively] (cf. Stewart, Taylor, & Baker 1997). The ANOVA on total ASI scores also revealed the expected highly signi®cant main e€ect of AS group [F (2, 87)=379.76, P< 0.001]. However, no signi®cant AS group  gender interaction was observed [F (2, 87)=0.86, n.s.], showing that the e€ect of AS group on ASI total scores did not di€er across the two gender groups. Means (and S.D.s) on the demographic and total ASI score variables are displayed in Table 1 as a function of AS group, collapsed across gender group. To determine whether drinking behavior varied as a function of AS group and/or gender, scores on the three drinking behavior measures (typical weekly drinking frequency, typical drinking quantity, and yearly excessive drinking frequency) were submitted to a 32 (AS group  gender) multivariate analysis of variance (MANOVA). A multivariate main e€ect of gender was observed [F (3, 85)=3.88, P< 0.05]. Signi®cant univariate gender e€ects were observed for the typical weekly drinking frequency measure [F (1, 87)=6.74, P< 0.05], the typical drinking quantity measure [F (1, 87)=5.29, P< 0.05], and the yearly excessive drinking frequency measure

162

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

[F (1, 87)=10.33, P< 0.005]. Men scored higher than the women in each case [Ms (and S.D.s)=1.7 (1.5) vs. 1.3 (1.2) drinking occasions per week; 5.7 (3.8) vs. 4.1 (2.3) standard drinks per drinking occasion; and 62.5 (81.1) vs. 32.2 (54.5) excessive drinking episodes per year, respectively] (cf. Wilsnack, Vogeltanz, Wilsnack & Harris, 2000). A multivariate main e€ect was also present for AS group [F (6, 172)=4.07, P< 0.005] with signi®cant univariate AS group e€ects observed only for the typical weekly drinking frequency measure [F (2, 90)=4.07, P< 0.05] and the yearly excessive drinking frequency measure [F (2, 90)=2.91, P< 0.05] Means (and S.D.s) on the drinking behavior measures are displayed in Table 1 as a function of AS group, collapsed across gender. A series of one-tailed independent sample t-tests revealed that for the typical weekly drinking frequency measure, high AS students reported more frequent drinking than did the low AS [t (62)=2.85, P< 0.005] or the moderate AS [t (57)=1.68, P< 0.05] students Ð low and moderate AS students did not di€er [t (61)=0.99, n.s. Table 1]. For the yearly excessive drinking frequency measure, the high AS students reported more frequent excessive drinking than did the low AS students [t (62)=2.36, P< 0.05]; moderate AS students failed to di€er from either the high AS [t (57)=.87, n.s.] or the low AS [t (61)=1.62, n.s.] students in terms of yearly excessive drinking frequency (Table 1). The overall MANOVA also revealed a signi®cant multivariate AS groupgender interaction [F (6, 172)=2.69, P< 0.05]. At the univariate level, the interaction was signi®cant only for yearly Table 1 Demographic, anxiety sensitivity, and drinking behavior scores as a function of anxiety sensitivity groupa Anxiety sensitivity group

Demographics Age Gender Education Anxiety sensitivity ASI Score Drinking behavior Frequency Quantity Excess a

M S.D. Females (frequency) Males (frequency) M S.D. M S.D. M S.D. M S.D. M S.D.

Low (n=34)

Moderate (n=29)

High (n=30)

23.6 6.3 19 15 2.9 1.1

21.5 3.7 18 11 2.8 1.1

21.1 2.7 23 7 2.6 0.9

16.3a,b 1.1

32.0a,c 6.4

1.3e 1.3 5.5 3.3 45.8 62.5

2.0d,e 1.5 4.0 2.5 62.4f 82.4

5.5b,c 1.3 1.0d 1.1 4.5 3.0 23.3f 47.5

ASI, Anxiety Sensitivity Index (Peterson & Reiss, 1992). Frequency, number of drinking occasions per week; Quantity, number of standard alcoholic beverages per drinking occasion; Excess, excessive drinking episodes per year. Means with similar letters di€er signi®cantly from one another (P< 0.05 or better). Results are presented collapsed across gender.

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

163

excessive drinking frequency [F (2, 87)=4.20, P< 0.05]. Among the men, low, moderate and high AS groups obtained means (and S.D.s) of 32.1 (66.0), 53.8 (65.5) and 141.1 (89.9) excessive drinking occasions per year, respectively. Among the women, low, moderate, and high AS groups obtained means (and S.D.s) of 16.4 (25.2), 40.9 (61.9) and 38.4 (64.6) excessive drinking occasions per year, respectively. A series of one-tailed independent sample t-tests were conducted separately by gender. High AS men reported signi®cantly more frequent excessive drinking than did the low AS [t (20)=3.22, P< 0.005] or the moderate AS [t (16)=2.39, P< 0.05] men Ð low and moderate AS men did not di€er signi®cantly [t (24)=0.83, n.s.]. The pattern was slightly di€erent and less robust for the women: high AS women reported only marginally more frequent excessive drinking than did the low AS women [t (40)=1.40, P< 0.10] and failed to di€er signi®cantly from moderate AS women [t (39)= 0.12, n.s.]. Moderate AS women reported marginally more frequent excessive drinking than did the low AS women [t (35)=1.59, P< 0.10]. 3.2. Mediator analyses: total sample Mediator analyses (Baron & Kenny, 1986) were performed ®rst using the data from the entire sample of self-classi®ed drinkers (n=182). Drinking motives were only tested as mediators in the case of hypothesized relations between AS levels and usual weekly drinking frequency and yearly excessive drinking frequency, since usual weekly drinking quantity was unrelated to AS levels in either the extreme group comparisons (Table 1) or in a multiple regression using ASI scores to predict drinking quantity in the entire sample of drinkers (t= 0.60, n.s.; r= 0.04). One-tailed tests were used given that directional e€ects had been hypothesized a priori. The set of regressions testing the mediating role of drinking motives in explaining high AS individuals' increased typical weekly drinking frequency are summarized in Fig. 1a. First, typical weekly drinking frequency scores were regressed on ASI scores. ASI scores predicted increased weekly drinking frequency (t=2.19, P< 0.05; see Fig. 1a). Next, Coping, Conformity, Enhancement, and Social Motives scores were separately regressed on ASI scores. ASI scores predicted increased Coping-Motivated drinking (t=4.66, P< 0.001; Fig. 1a) and increased ConformityMotivated drinking (t=4.04, P< 0.001; Fig. 1a). However, ASI scores did not predict Enhancement Motives (t=0.06, n.s.; r=0.00) or Social Motives (t=1.02, n.s., r=0.08) scores. Thus, only Coping and Conformity Motives could be evaluated further as potential mediator variables. Next, typical weekly drinking frequency scores were regressed on ASI, Coping Motives, and Conformity Motives scores simultaneously. Examination of the univariate e€ects of each variable indicated that Coping Motives predicted increased typical weekly drinking frequency (t=4.11, P< 0.001) and Conformity Motives were a signi®cant, independent predictor of increased typical weekly drinking frequency (t=1.84, P< 0.05), whereas ASI scores no longer predicted increased typical weekly drinking frequency (t=0.25, n.s.) after accounting for the in¯uences of Coping and Conformity Motives (Fig. 1a). The set of regressions testing the mediating role of drinking motives in explaining high AS individuals' increased yearly excessive drinking frequency are summarized in Fig. 1b. Yearly excessive drinking frequency scores were ®rst regressed on ASI scores. ASI scores predicted increased yearly excessive drinking frequency (t=1.61, P< 0.05; Fig. 1b). Next, Coping, Conformity, Enhancement, and Social Motives scores were separately regressed on ASI scores. As was the case earlier, only Coping Motives and Conformity Motives could be evaluated further as

164

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

potential mediator variables (Fig. 1b). Next, yearly excessive drinking frequency scores were regressed on ASI scores, Coping Motives scores, and Conformity Motives scores simultaneously. Coping Motives predicted increased yearly excessive drinking frequency (t=4.82, P< 0.001) and Conformity Motives were a signi®cant independent predictor of increased yearly excessive drinking frequency (t=2.08, P< 0.05), whereas ASI scores were no longer a signi®cant predictor of increased yearly excessive drinking frequency (t=-0.60, n.s.) after accounting for the in¯uences of Coping and Conformity Motives (Fig. 1b). 3.3. Mediator analyses in women versus men The set of regressions testing the mediating role of drinking motives in explaining high AS individuals' increased drinking behavior were then tested separately within each gender group. First, typical weekly drinking frequency, typical quantity, and yearly excessive drinking scores were separately regressed on ASI scores in the women. ASI scores predicted increased weekly drinking frequency among women (t=1.87, P< 0.05; r=0.18) but failed to predict women's drinking quantity (t= 0.62, n.s., r= 0.06) or excessive drinking (t=0.86, n.s., r=0.08). Thus,

Fig. 1. Summary of mediating roles of negative reinforcement drinking motives in explaining relation between anxiety sensitivity levels and drinking behavior with (a) usual weekly drinking frequency and (b) yearly excessive drinking episodes, as the criterion variables.**P< 0.001; *P< 0.05 (one-tailed tests). Partial correlation coecients are displayed in parentheses. Anxiety Sensitivity assessed with the Anxiety Sensitivity Index (Peterson & Reiss, 1992); Coping and Conformity drinking motives assessed with the Revised Drinking Motives Questionnaire (Cooper, 1994); Drinking Behavior assessed with an author-compiled questionnaire (cf. Stewart et al., 1995). Partial correlations between Anxiety Sensitivity and drinking behavior criterion measures control for levels of Coping and Conformity Motives. Partial correlations between Coping Motives and drinking behavior criterion measures control for levels of Anxiety Sensitivity and Conformity Motives. Partial correlations between Conformity Motives and drinking behavior criterion measures control for levels of Anxiety Sensitivity and Coping Motives. n=182 drinkers (collapsed across gender).

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

165

mediator analyses could only be pursued further for weekly drinking frequency for the women. Next, Coping, Conformity, Enhancement, and Social Motives scores were separately regressed on ASI scores in the women. ASI scores predicted increased Coping-Motivated drinking (t=5.00, P< 0.001; r=0.44) and increased Conformity-Motivated drinking (t=2.82, P< 0.005; r=0.26) in the women. However, women's ASI scores did not predict Enhancement Motives (t=0.09, n.s.; r=0.01) or Social Motives (t=0.46, n.s., r=0.04) scores. Thus, only Coping and Conformity Motives could be evaluated further as potential mediator variables in the women. Next, typical weekly drinking frequency scores were regressed on ASI, Coping Motives, and Conformity Motives scores simultaneously. Examination of the univariate e€ects of each variable indicated that Coping Motives (t=1.98, P< 0.05; partial r=0.19) but not Conformity motives (t=1.19, n.s.; partial r=0.02), predicted increased typical weekly drinking frequency among the women. In contrast, women's ASI scores no longer predicted increased typical weekly drinking frequency (t=0.23, n.s.; partial r=0.07) after accounting for the in¯uences of Coping Motives. This result supports Coping Motives as a mediator of the relation between high AS and increased drinking behavior among women. Next, typical weekly drinking frequency, drinking quantity, and yearly excessive drinking scores were separately regressed on ASI scores in the men. ASI scores predicted increased weekly drinking frequency scores (t=2.08, P< 0.05; r=0.24) and yearly excessive drinking frequency (t=2.04, P< 0.05; r=24), but not typical drinking quantity (t=0.16, n.s.; r=0.02) among the men. Thus, only typical frequency and yearly excessive drinking could be pursued further in mediator analyses for the men. Next, Coping, Conformity, Enhancement, and Social Motives scores were separately regressed on ASI scores in the men. ASI scores predicted increased Conformity Motivated drinking in the men (t=3.63, P< 0.001; r=0.40). However, men's ASI scores did not predict Enhancement Motives (t=0.51, n.s.; r=0.06) or Social Motives (t=1.63, n.s.; r=0.19) scores among the men. Moreover, unlike the ®ndings with the women, men's ASI scores did not predict Coping Motives scores (t=1.46, n.s.; r=0.17). Thus, only Conformity Motives could be evaluated further as a potential mediator variable in the men. Typical weekly drinking frequency scores were regressed on ASI and Conformity Motives scores simultaneously. Examination of the univariate e€ects of each variable indicated that Conformity Motives predicted increased typical weekly drinking frequency (t=2.41, P< 0.01; partial r=0.28) among the men whereas men's ASI scores no longer predicted increased typical weekly drinking frequency after accounting for the in¯uences of Conformity Motives (t=1.02, n.s.; partial r=0.12). When ASI and Conformity Motives scores were entered simultaneously to predict yearly excessive drinking frequency, only Conformity Motives scores (t=1.54, P=0.06; r=0.18) but not ASI scores (t=1.28, n.s.; partial r=0.15) served as a marginally signi®cant predictor at the univariate level. Taken together, these results support Conformity Motives as a mediator of the relation between high AS and increased drinking behavior in men. 3.4. Relations of AS lower-order components to aspects of drinking behavior In order to examine the relations between particular lower-order AS components and the various aspects of drinking behavior in our sample, we ®rst conducted principal components analysis on the ASI item scores of the total sample of drinkers in order to ensure that we had a comparable factor structure in the present sample to that observed in previous studies (see review by

166

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

Zinbarg et al., 1999). A three-factor solution (eigenvalues >1.0=6.2, 1.6, and 1.3) accounted for 56.8% of the variance and produced a clearly interpretable factor pattern matrix following oblique (Oblimin) rotation to allow for intercorrelation between factors (cf. Stewart, Taylor & Baker 1997). The solution showed excellent simple structure (Thurstone, 1947); de®ning salient loadings as >0.35 there were no hyperplane items and only one complex item (i.e. item 13). The ®rst factor showed non-complex salient loadings from items 4, 6, 8, 10, 9, 3, 14, and 11 and was interpreted as re¯ecting AS physical concerns (e.g. ``When my heart is beating rapidly, I worry that I might have a heart attack''). The second factor showed non-complex salient loadings from items 2, 15, 12, 16, and 7 and was interpreted as re¯ecting AS psychological concerns (e.g. ``When I cannot keep my mind on a task, I worry that I might be going crazy''). The third factor showed noncomplex salient loadings from items 5 and 1 and was interpreted as re¯ecting AS social concerns (``It is important to me not to appear nervous''). Inter-correlations between the factors ranged from r=0.21 (psychological with social concerns) to 0.42 (physical with psychological concerns; all P< 0.005). Given the similarity of this factor structure with those previously reported for the ASI (see Zinbarg et al., 1999), we calculated factor scores on each of the three factors for each participant using the regression method (cf. Samoluk & Stewart, 1998). Next, physical, psychological, and social concern factor scores were correlated with each of the three drinking behavior variables in the total sample of drinkers (n=182). The relevant bivariate correlations are displayed in Table 2. None of the three ASI factor scores were signi®cantly related to typical drinking quantity. Typical weekly drinking frequency was signi®cantly positively correlated with scores on the social concern and psychological concern ASI factors (Table 2). Yearly excessive drinking frequency was related only to social concern factor scores (Table 2). Given the inter-correlation among lower-order AS factor scores, it was important to examine the unique contributions of each AS component to the prediction of typical weekly drinking frequency and yearly excessive drinking frequency. Thus, for each drinking behavior outcome, a multiple regression analysis was performed with each of the three AS factor score predictors being entered simultaneously. Together the three AS components signi®cantly predicted weekly drinking frequency [F (3, 178)=4.24, P< 0.01]. Only AS social concerns proved a signi®cant predictor at the univariate level (t=2.98, P< 0.005; partial r=0.22). Similarly, the Table 2 Bivariate correlations between lower-order anxiety sensitivity components and drinking behavior measuresa Drinking Behavior Measure

Anxiety sensitivity component Physical concerns

Frequency Quantity Excess

0.05 0.05 0.03

Psychological concerns 0.14* 0.05 0.10

Social concerns 0.24*** 0.02 0.19**

a Physical, Psychological, and Social Concern scores are obliquely rotated factor scores derived via the regression method following principal components analysis of Anxiety Sensitivity Index (Peterson & Reiss, 1992) item scores. Frequency, number of drinking occasions per week; Quantity, number of standard alcoholic beverages per drinking occasion; Excess, excessive drinking episodes per year. Results are presented collapsed across gender. *P< 0.05 **P< 0.005 ***P< 0.001 (one-tailed tests).

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

167

three AS components as a block signi®cantly predicted yearly excessive drinking [F (3, 178)=2.59, P< 0.05]. Again, only AS social concerns proved a signi®cant predictor at the univariate level (t=2.44, P< 0.05; partial r=0.18). 4. Discussion Our results provide further support for the contention that elevated AS levels are associated with increased drinking behavior (cf. Cox et al., 1993; Stewart et al., 1995). Speci®cally, we found that the elevated drinking levels displayed by our sample of high AS young adults, relative to moderate and low AS young adults, was apparent for typical weekly drinking frequency, but not for drinking quantity. Nonetheless, high AS persons did report drinking to excess more frequently than others (cf. Stewart et al. 1995). Thus, similar to panic disorder patients (Lehman, Brown, & Barlow, 1998), the quantities of alcohol being more frequently consumed by high AS individuals appear likely to be in a clinically-signi®cant range (Wechsler et al., 1994). In fact, some researchers advocate use of a de®nition of problem drinking involving drinking to excess more than 36 times per year (see review in DeCourville & Sadava, 1997). Examination of the data in Table 1 indicates that the average high AS student surpasses this problem drinking cuto€, whereas the average low AS student does not. There are several possible explanations for the observed results regarding the associations of AS levels with frequency vs. quantity of alcohol use. First, although the emotional reactivity dampening e€ects of alcohol are indeed dose dependent, high AS young adults obtain a signi®cant ``dampening e€ect'' at relatively lower alcohol doses (MacDonald et al., 2000). Thus, high AS persons may experience signi®cant alcohol-induced anxiety reduction even from levels of drinking producing only mild-to-moderate levels of intoxication. Second, alcohol may serve as a ``safety signal'' for high AS individuals, that identi®es: (1) the o€set of elevated anxiety, and/or (2) the absence of the onset of anxiety sensations (cf. Seligman & Binik, 1977). Just as panic disorder patients believe that alcohol can prevent or reduce the occurrence of panic attacks (Bibb & Chambless, 1986; Chambless, Cherney, Caputo, & Rheinstein, 1987), high AS undergraduates may hold increased tension-reduction ``alcohol outcome expectancies'' (Goldman, Del Boca, & Darkes, 1999) which may contribute to their more frequent drinking (cf. Stewart, Samoluk, & MacDonald, 1999). Third, an increased drinking quantity might be expected only in high AS drinkers that have greater experience with the frequent use of alcohol to cope. Such individuals may develop tolerance to the anxiolytic properties of alcohol, and thus require higher levels of alcohol to achieve anxiety-reducing e€ects over time (Stewart, Samoluk, & MacDonald, 1999). Of course, these potential explanations warrant future empirical evaluation. One gender moderator e€ect was observed in the relation of AS levels to drinking behavior. Speci®cally, the relation of high AS to yearly excessive drinking frequency was stronger among the men than among the women. This is consistent with previous ®ndings regarding gender differences in the relation of AS to increased drinking behavior among panic disorder patients (Cox et al., 1993). However, the relation between AS levels and increased excessive drinking among university women in the present study was less robust than that observed by Stewart et al. (1995). It should be cautioned that the gender moderator e€ect for excessive drinking might be secondary to the way in which excessive drinking was assessed in the present study. Since we did not collect

168

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

individual body weight data, we were unable to calculate excessive drinking frequency adjusted for participants' body weights. Therefore, we may have overestimated the number of drinks that might be considered risky or ``excessive'' for lighter weight individuals. Given the lower average body weights of women, many women may require less alcohol to reach an ``excessive'' threshold then most men. In contrast to the present method of de®ning excessive drinking as the frequency of drinking four or more drinks in a single occasion, Stewart et al. de®ned excessive drinking as the frequency of drinking to the point of legal intoxication. This measure was calculated individually for each participant, adjusted for body weight (cf. Conrod, Stewart, & Pihl, 1997). We also found strong support for the mediational role of Coping Motives, and additional support for Conformity Motives as a reliable independent mediator, in explaining the relation between high AS and increased drinking behavior in the total sample of student drinkers. Speci®cally, AS failed to predict either increased weekly drinking frequency or yearly excessive drinking episodes after accounting for the in¯uences of Coping and Conformity Motives. This suggests that high AS individuals' greater drinking behavior can be at least partially explained by their tendency to drink to control aversive states (cf. Baron & Kenny, 1986). This ®nding may have potentially important clinical implications. For example, future research might investigate the role of AS in predicting relapse among problem drinkers during the early stages of abstinence attempts when anxiety-related distress is likely to be greatest. We also found interesting gender di€erences in the speci®c drinking motives that mediated the association between AS and increased drinking behavior. The fact that Coping Motives but not Conformity Motives served as a signi®cant mediator among the women may indicate that the increased drinking frequency of high AS women is primarily motivated by the desire to avoid/ escape unwanted internal states. This ®nding points to the greater importance of Coping Motives in understanding high AS women's drinking behavior relative to high AS men (cf. Stewart, Karp et al., 1997; Stewart & Zeitlin, 1995). In contrast, the fact that Conformity Motives served as the signi®cant mediator among the men suggests that avoidance/escape of aversive external consequences (e.g. social embarrassment) may motivate the increased drinking behavior of high AS men. This is likely related to the relatively high levels of AS ``social concerns'' among men (e.g. Stewart, Karp et al., 1997; Stewart et al., 1998). Increased attention should be devoted to gender di€erences in the underlying motivations for alcohol use in high AS individuals in future research. A ®nal purpose of the present study was to examine relations between speci®c lower-order AS components and increased drinking behavior. Consistent with the ®ndings of the only previous study to examine this issue (i.e. McWilliams & Asmundson, 1999), we found AS social concerns to be the only independent lower-order component of the ASI that was associated with increased drinking behavior. This pattern again provides increased understanding of the motivations for increased alcohol use among high AS individuals. Speci®cally, their desire to dampen publicly observable anxiety symptoms may be crucial in explaining high AS individuals' more frequent alcohol use and increased excessive drinking. Moreover, this ®nding points to the potential importance of AS to understanding the common co-morbidity of social phobia and alcoholrelated disorders (Kushner, Abrams, & Borchardt, 2000), since the AS social concerns component has been shown to be the most important AS component in social phobia (Zinbarg et al., 1997). Global AS levels, and AS social concerns in particular, were found to account for a relatively small but signi®cant proportion of variance in two of the three drinking behavior indices examined (cf. McWilliams & Asmundson, 1999). Nonetheless, because there often are multiple determinants

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

169

to drinking behavior, any one dispositional factor such as the ASI may only have weak associations with drinking behavior (Schuckit & Hesselbrock, 1994). Investigation of other dispositional and contextual factors will therefore be an important endeavor. It would be interesting for future research to test McNally's (1996) prediction that the interaction between AS and trait anxiety should be strongly related to alcohol use/abuse, and to evaluate the potential mediating role of the negative reinforcement drinking motives in accounting for this hypothesized relation, given that AS and trait anxiety have been shown to independently predict Coping Motives (Stewart & Zeitlin, 1995). The present study has a number of limitations. First, the correlational and cross-sectional nature of our results precludes causal interpretations of the observed relations between AS, drinking motives, and drinking behavior. Second, the relatively homogeneous demographic composition of the sample may limit the generalizability of the present ®ndings to ethnic minorities (e.g. McNeil, Kee, & Zvolensky, 1999; Zvolensky, McNeil, Porter, & Stewart, 2001). Third, we did not include a measure of drinking problems (cf. Conrod et al., 1998; Cooper et al., 1995); thus, speculations regarding the potential links between the present ®ndings and the development of problem drinking remain tentative. Fourth, the Coping Motives subscale of the DMQ-R fails to distinguish between reasons for drinking involving controlling anxious emotions and those involving controlling arousal-related bodily states. Determining which of these two types of coping motives mediates high AS individuals' increased drinking behavior will be an important future endeavor. Finally, we relied solely on self-report methods; using multiple methods of assessment (e.g. behavioral observation, psychophysiological measures) in controlled environments would be bene®cial in the future (e.g. Eifert, Forsyth, Zvolensky, & Lejuez, 1999; Samoluk, Stewart, Sweet, & MacDonald, 1999). Taken together, our results suggest elevated AS levels may increase the likelihood of drinking for negative-reinforcement reasons (i.e. Coping or Conformity Motives), which in turn may contribute to more frequent drinking behavior, including more frequent drinking to excess. Although these data are correlational, it is possible that elevated AS may place individuals at greater risk for drinking problems by way of its association with risky drinking motives and greater drinking behavior. Acknowledgements This study was supported, in part, by grants from the Alcoholic Beverage Medical Research Foundation (ABMRF) and Social Sciences and Humanities Research Council of Canada (SSHRC) awarded to the ®rst author, by a grant-in-aid-of-research from the Sigma Xi Research Society awarded to the second author, and by a research grant from the Eberly Family Trust awarded to the third author. The authors would like to thank Brent Conrad, Paula MacPherson, and Heather Lee Loughlin for their research assistance at the Dalhousie University site, and Courtney Churches for her research assistance at the West Virginia University site. References Ball, S. G., Otto, M. W., Pollack, M. H., Uccello, R., & Rosenbaum, J. F. (1995). Di€erentiating social phobia and panic disorder: A test of core beliefs. Cognitive Therapy and Research, 19, 473±482.

170

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173±1182. Bibb, J., & Chambless, D. L. (1986). Alcohol use and abuse among diagnosed agoraphobics. Behaviour Research and Therapy, 24, 49±58. Carrigan, G., Samoluk, S. B., & Stewart, S. H. (1998). Examination of the short form of the Inventory of Drinking Situations (IDS-42) in a young adult university student sample. Behavior Research and Therapy, 36, 789±807. Chambless, D. L., Cherney, J., Caputo, G. C., & Rheinstein, B. J. (1987). Anxiety disorders and alcoholism: a study with inpatient alcoholics. Journal of Anxiety Disorders, 1, 29±40. Conrod, P. J., Pihl, R. O., & Vassileva, J. (1998). Di€erential sensitivity to alcohol reinforcement in groups of men at risk for distinct alcoholic syndromes. Alcoholism: Clinical and Experimental Research, 22, 585±597. Conrod, P. J., Stewart, S. H., & Pihl, R. O. (1997). Validation of a measure of excessive drinking: Frequency per year that BAL exceeds 0.08%. Substance Use and Misuse, 32, 587±607. Cooper, M. L. (1994). Motivations for alcohol use among adolescents: development and validation of a four-factor model. Psychological Assessment, 6, 117±128. Cooper, M. L., Frone, M. R., Russell, M., & Mudar, P. (1995). Drinking to regulate positive and negative emotions: A motivational model of alcohol use. Journal of Personality and Social Psychology, 69, 990±1005. Cooper, M. L., Russell, M., Skinner, J. B., & Windle, M. (1992). Development and validation of a three-dimensional measure of drinking motives. Psychological Assessment, 4, 123±132. Cox, B. J., Swinson, R. P., Shulman, I. D., Kuch, K., & Reichman, J. T. (1993). Gender e€ects and alcohol use in panic disorder with agoraphobia. Behaviour Research and Therapy, 31, 413±416. Cox, W. M., & Klinger, E. (1988). A motivational model of alcohol use. Journal of Abnormal Psychology, 97, 168±180. DeCourville, N. H., & Sadava, S. W. (1997). The structure of problem drinking in adulthood: a con®rmatory approach. Journal of Studies on Alcohol, 58, 146±154. Eifert, G. H., Forsyth, J. P., Zvolensky, M. J., & Lejuez, C. W. (1999). Moving from the laboratory to the real world and back again: Increasing the relevance of laboratory examinations of anxiety sensitivity. Behavior Therapy, 30, 273±283. Goldman, M. S., Del Boca, F. K., & Darkes, J. (1999). Alcohol expectancy theory: the application of cognitive neuroscience. In K. E. Leonard, & H. T. Blane, Psychological theories of drinking and alcoholism, Second Edition (pp. 203±246). New York: Guilford. Kushner, M. B., Abrams, K., & Borchardt, C. (2000). The relationship between anxiety disorders and alcohol use disorders: A review of major perspectives and ®ndings. Clinical Psychology Review, 20, 149±171. Lehman, C. L., Brown, T. A., & Barlow, D. H. (1998). E€ects of cognitive-behavioral treatment for panic disorder with agoraphobia on concurrent alcohol abuse. Behavior Therapy, 29, 423±433. MacDonald, A. B., Baker, J. M., Stewart, S. H., & Skinner, M. (2000). The e€ects of alcohol on response to hyperventilation in high and low anxiety sensitive participants. Alcoholism: Clinical and Experimental Research, 24(11), 1656±1665. Maller, R. G., & Reiss, S. (1992). Anxiety sensitivity in 1984 and panic attacks in 1987. Journal of Anxiety Disorders, 6, 241±247. McNally, R. J. (1996). Anxiety sensitivity is distinguishable from trait anxiety. In R. M. Rapee, Current controversies in the anxiety disorders (pp. 214±227). New York: Guilford. McNeil, D. W., Kee, M., & Zvolensky, M. J. (1999). Culturally-related anxiety and ethnic identity in Navajo college students. Cultural Diversity and Ethnic Minority Psychology, 5, 56±64. McWilliams, L. A., & Asmundson, G. A. (1999). Alcohol consumption in university women: a second look at the role of anxiety sensitivity. Depression and Anxiety, 10, 125±128. McWilliams, L. A., Stewart, S. H., & MacPherson, P. S. R. (2000). Does the social concerns component of the Anxiety Sensitivity Index belong to the domain of anxiety sensitivity or the domain of negative evaluation sensitivity?. Behaviour Research and Therapy, 38, 985±992. Peterson, R. A., & Reiss, S. (1992). Anxiety sensitivity index manual (2nd ed.). Worthington, OH: International Diagnostic Systems. Samoluk, S. B., & Stewart, S. H. (1998). Anxiety sensitivity and situation-speci®c drinking. Journal of Anxiety Disorders, 12, 407±419.

S.H. Stewart et al. / Personality and Individual Di€erences 31 (2001) 157±171

171

Samoluk, S. B., Stewart, S. H., Sweet, S., & MacDonald, A. B. (1999). Anxiety sensitivity and social aliation as determinants of alcohol consumption. Behavior Therapy, 30, 285±303. Schmidt, N. B., Lerew, D. R., & Jackson, R. J. (1997). The role of anxiety sensitivity in the pathogenesis of panic: Prospective evaluation of spontaneous panic attacks during acute stress. Journal of Abnormal Psychology, 106, 355± 364. Schuckit, M. A., & Hesselbrock, V. (1994). Alcohol dependence and anxiety disorders: what is the relationship?. American Journal of Psychiatry, 151, 1723±1734. Seligman, M. E. P., & Binik, Y. (1977). The safety signal hypothesis. In H. Davis, & H. Hurwitz, Pavolvian-operant interactions (pp. 165±187). Hillsdale, NJ: Erlbaum. Sobell, L. C., & Sobell, M. C. (1990). Self-report issues in addictive behaviors [Special issue]. Behavioral Assessment, 12(1). Stewart, S. H., Angelopoulos, M., Baker, J. M., & Boland, F. J. (2000). Relations between dietary restraint and patterns of alcohol use in young adult women. Psychology of Addictive Behaviors, 14, 77±82. Stewart, S. H., Conrod, P. J., Gignac, M. L., & Pihl, R. O. (1998). Selective processing biases in anxiety sensitive men and women. Cognition and Emotion, 12, 105±133. Stewart, S. H., & Devine, H. (2000). Relations between personality and drinking motives in young adults. Personality and Individual Di€erences, 29, 495±511. Stewart, S. H., Karp, J., Pihl, R. O., & Peterson, R. A. (1997). Anxiety sensitivity and self-reported reasons for drug use. Journal of Substance Abuse, 9, 223±240. Stewart, S. H., Knize, K., & Pihl, R. O. (1992). Anxiety sensitivity and dependency in clinical and non-clinical panickers and controls. Journal of Anxiety Disorders, 6, 119±131. Stewart, S. H., Loughlin, H. L., & Rhyno, E. (2001). Internal drinking motives mediate personality domain Ð drinking relations in young adults. Personality and Individual Di€erences, 30, 271±286. Stewart, S. H., Peterson, J. B., & Pihl, R. O. (1995). Anxiety sensitivity and self-reported alcohol consumption rates in university women. Journal of Anxiety Disorders, 9, 283±292. Stewart, S. H., & Pihl, R. O. (1994). The e€ects of alcohol administration on psychophysiological and subjectiveemotional responses to aversive stimulation in anxiety sensitive women. Psychology of Addictive Behaviors, 8, 29±42. Stewart, S. H., Samoluk, S. B., & MacDonald, A. B. (1999). Anxiety sensitivity and substance use and abuse. In S. Taylor, Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety (pp. 287±319). Mahwah, NJ: Erlbaum. Stewart, S. H., Taylor, S., & Baker, J. M. (1997). Gender di€erences in dimensions of anxiety sensitivity. Journal of Anxiety Disorders, 11, 179±200. Stewart, S. H., & Zeitlin, S. B. (1995). Anxiety sensitivity and alcohol use motives. Journal of Anxiety Disorders, 9, 229± 240. Stewart, S. H., Zvolensky, M. J., & Eifert, G. H. (1999b). The relations of anxiety sensitivity, experiential avoidance, and alexithymic coping to young adults' motivations for drinking [Summary]. Proceedings of the 33rd Annual Convention of the Association for Advancement of Behavior Therapy (CD-ROM). Thurstone, L. L. (1947). Multiple factor analysis. Chicago, IL: The University of Chicago Press. Vogel-Sprott, M. (1983). Response measures of social drinking: Research implications and applications. Journal of Studies on Alcohol, 44, 817±836. Wechsler, H., Davenport, A., Dowdall, G., Moeykens, B., & Castillo, S. (1994). Health and behavioral consequences of binge drinking in college: a national survey of students at 140 campuses. Journal of the American Medical Association, 272, 1672±1677. Wilsnack, R. W., Vogeltanz, N. D., Wilsnack, S. C., & Harris, T. R. (2000). Gender di€erences in alcohol consumption and adverse drinking consequences: Cross-cultural patterns. Addiction, 95, 251±265. Zinbarg, R. E., Barlow, D. H., & Brown, T. A. (1997). Hierarchical structure and general factor saturation of the Anxiety Sensitivity Index: evidence and implications. Psychological Assessment, 9, 277±284. Zinbarg, R. E., Mohlman, J., & Hong, N. N. (1999). Dimensions of anxiety sensitivity. In S. Taylor, Anxiety sensitivity: Theory, research, and treatment of the fear of anxiety (pp. 83±114). Mahwah, NJ: Erlbaum. Zvolensky, M. J., McNeil, D. W., Porter, C. A., & Stewart, S. H. (2001). Assessment of anxiety sensitivity in American Indian and Alaska Natives. Behaviour Research and Therapy, 39, 477±493.