Assessing Self-Efficacy to Reduce One's Drinking: Further Evaluation of the Alcohol Reduction Strategies-Current Confidence Questionnaire

Abstract

Aims: To evaluate the psychometric properties of a previously published questionnaire designed to assess young drinkers’ self-efficacy to employ 31 cognitive-behavioral alcohol reduction strategies. Methods: Undergraduates (n = 353) recruited from a large Midwestern university completed the previously published Alcohol Reduction Strategies-Current Confidence questionnaire (and other measures) for a self-selected heavy drinking setting. Results: Item loadings from a principal components analysis, a high internal consistency reliability coefficient, and a moderate mean inter-item correlation suggested that all 31 items comprised a single scale. Correlations of questionnaire scores with selected aspects of drinking history and personality provided support for criterion and discriminant validity, respectively. Women reported higher current confidence to use these strategies than did men, but current confidence did not vary as a function of recent binge status. Conclusion: Given this further demonstration of its psychometric qualities, this questionnaire holds promise as a clinical tool to identify clients who lack confidence in their ability to employ cognitive-behavioral coping strategies to reduce their drinking.

INTRODUCTION

Although there are multiple benefits of young people employing alcohol-reduction strategies to reduce the quantity, frequency and speed of their drinking, many university students do not use such strategies consistently to moderate their drinking (Benton et al., 2004; Martens et al., 2007b). Students may not employ alcohol-reduction strategies if they perceive it as socially unacceptable to do so or if they seek the experience of intense intoxication that follows rapid consumption of large amounts of alcohol. Another reason these strategies may not be employed is that some young people have limited self-efficacy or self-confidence that they could utilize specific drinking self-control strategies in certain situations (Ray et al., 2009).

Several research teams have published self-efficacy questionnaires designed to assess drinkers’ confidence to abstain or moderate their drinking in different contexts (Annis and Graham, 1988; DiClemente et al., 1994; O'Hare, 2001; Sitharthan et al., 2003; Oei et al., 2005; Young et al., 2007), and other investigators have published questionnaires that measure past use of specific drinking-reduction and harm-reduction skills (e.g. Martens et al., 2007a; Sugarman and Carey, 2007). However, none of these instruments were designed to assess self-efficacy to employ specific strategies to restrain one's drinking. Therefore, Bonar et al. (2011) developed the Alcohol Reduction Strategies-Current Confidence questionnaire (ARS-CC), a self-administered instrument that asks respondents to rate their confidence to utilize each of 31 cognitive-behavioral alcohol reduction strategies [e.g. ‘Avoid adding more alcohol to a drink you have not finished’, ‘Set down your drink between each sip’, ‘Avoid drinking in rounds (e.g. taking turns buying drinks for a group’)] in drinker-selected or clinician-specified drinking contexts.

Bonar et al.'s evaluation of the ARS-CC found that undergraduates reported higher mean self-efficacy when they imagined drinking in their own dorm/apartment than when drinking in bars or at parties; women had higher mean self-efficacy scores than men; and drinkers who engaged in three or more binge episodes in the previous 2 weeks had lower mean self-efficacy scores than those who reported fewer episodes in the same time period. In addition, the ARS-CC had excellent internal consistency reliability, a moderate mean inter-item correlation and good 1-week test-retest reliability. Also, ARS-CC scores correlated significantly with several measures of drinking history.

As a follow-up to this initial investigation by Bonar et al. (2011), we designed the present study to evaluate further the psychometric properties of the ARS-CC when administered to a different sample of binge-drinking university students. Specifically, we evaluated its factor structure and internal consistency reliability, its criterion validity with several aspects of drinking history (e.g. quantity/frequency, consequences), and its discriminant validity with sensation seeking, alcohol outcome expectancies and self-efficacy to achieve non-drinking health outcomes. Finally, because one's reported self-efficacy to employ drinking self-control skills could be influenced by social desirability bias, we also evaluated the association of ARS-CC scores with measures of impression management and self-deception.

METHODS

Procedure and participants

Upon receiving approval for the replication study from our institutional review board, students were recruited from introductory and upper-level psychology courses at a large public Midwestern university (enrollment ∼20,000 students). Potential participants were invited to click a web link in the recruitment notice that opened an informed consent page followed by the questionnaires described below. To be eligible for inclusion in the database, respondents had to be between 18 and 29 years of age and to have reported consuming at least five drinks (if male) or at least four drinks (if female) on at least one occasion in the previous month. Participants who completed the materials received research credit points.

We recruited 358 undergraduates who met these criteria. We excluded from further analysis those four students who selected more than one preferred 5+/4+ location and the one student who reported the heaviest consumption in a unique setting (i.e. parent's home). Of the remaining 353 participants, 188 engaged in 5+/4+ binge episodes most often at a house/fraternity/sorority party, 90 did so most often in their own dorm room or own apartment, and 75 did so most often in a bar/club/restaurant. Frequency counts revealed that 62% of participants were female, and 80% indicated their ethnicity was White/European American, both of which are consistent with the proportions of women and Caucasian students enrolled at this university. Other demographic and drinking characteristics are reported in Table 1.

Measures

Alcohol Reduction Strategies-Current Confidence

This computer-based questionnaire asked each respondent to rate his/her current confidence to employ each of 31 different drinking-reduction self-control skills [see Table 2 for a list of strategies and means (SDs) for this sample] while drinking in the specific location where he/she reported most often consuming 5+/4+ drinks in a row. Because situational specificity theory proposes that behavior often varies across contexts, we asked participants to identify the one situation in which they consumed 5+/4+ drinks most often and to rate their use of the strategies in that context. Each item had the following five response options presented without numerical notation: not at all confident, a little confident, moderately confident, very confident and completely confident (coded 1–5, respectively for statistical analyses). In addition, participants were also able to indicate if a listed strategy did not apply to them (coded as missing data for further analyses).

Short Rutgers Alcohol Problem Index

We used the 16-item short form (Earleywine et al., 2008) of the original 23-item Rutgers Alcohol Problem Index (RAPI) (White and Labouvie, 1989) to assess consequences of participants’ alcohol use over the past 3 years. To increase clarity of phrasing, we shortened the one item that asks about efforts to control one's drinking. Earleywine et al. (2008) reported that the S-RAPI had good internal consistency reliability and correlated highly with the full RAPI. Cronbach's α in our sample was 0.88.

Comprehensive Effects of Alcohol

The 38-item Comprehensive Effects of Alcohol (CEOA) questionnaire (Fromme et al., 1993) was used to assess participants’ positive and negative outcome expectancies of being under the influence of alcohol. Previous research supported the internal consistency, test-retest reliability and construct validity of the CEOA and its two main subscales (Fromme et al., 1993; Ham et al., 2005). Cronbach's a in our sample was 0.90 for the positive expectancies subscale and 0.88 for the negative expectancies subscale.

Alcohol Use Disorders Identification Test

We used the 10-item Alcohol Use Disorders Identification Test (AUDIT) to assess whether participants engaged in problem drinking (Saunders et al., 1993). Previous research has supported the reliability and concurrent validity of the AUDIT (Reinert and Allen, 2002; Selin, 2003). Cronbach's α in our sample was 0.79.

Health Self-Efficacy Scale

We included this scale to measure self-efficacy to engage in five health-promoting activities over the next year (Grembowski et al., 1993). Using an 11-point scale (0 = not at all sure to 10 = very sure), participants were asked to rate their self-efficacy to control their weight, eat less fat, exercise regularly, not drink heavily and not smoke. Cronbach's α in our sample was 0.61.

Brief Sensation Seeking Scale

The 8-item Brief Sensation Seeking Scale (Hoyle et al., 2002) was used to assess participants’ tendencies to seek out varied and novel situations, a characteristic that has been related to excessive drinking (Hittner and Swickert, 2006). Cronbach's α in our sample was 0.81.

Balanced Inventory of Desirable Responding

The 40-item balanced inventory of desirable responding (BIDR) (Paulhus, 1984, 1991) yields two subscales: self-deception (unintentionally portraying oneself in a favorable light) and impression management (intentionally portraying oneself positively in order to be perceived favorably by others). Paulhus (1991) reported the subscales demonstrated good internal consistency and test-retest reliability, and several aspects of validity. Cronbach's α in our sample was 0.72 for the self-deception subscale and 0.69 for the impression management subscale.

Background Questionnaire

This questionnaire was developed to assess basic demographic information and drinking history. These questions included items to assess typical number of drinks per week, typical number of drinks per drinking occasion, perceived ease of abstaining in the next 30 days and level of perceived control over one's drinking.

RESULTS

Psychometric properties of the ARS-CC

Firstly, we conducted a principal components analysis (promax rotation). The results suggested six clusters of items, but our examination of the eigenvalues (11.70, 2.29, 1.50, 1.35, 1.22 and 1.09) and proportions of variance accounted for by each component (37.7, 7.4, 4.8, 4.3, 3.9 and 3.5%), led us to interpret the 31 items as comprising a single scale. This interpretation was further supported by the finding that 30 of the 31 items loaded most highly (loadings ranged from 0.49 to 0.79) on the first component; furthermore, the one item (‘Eat a meal before starting to drink’) that loaded more highly on the second component also cross-loaded on the initial component. We also evaluated the alpha coefficient and mean inter-item correlation across all 31 items. Internal reliability was notably high (α = 0.96). As recommended for a targeted construct (Clark and Watson, 1995) such as the one assessed by this questionnaire, the mean inter-item correlation was moderate (average r = 0.43).

Correlations of ARS-CC scores with drinking characteristics

We also conducted Pearson's product-moment correlations to examine the relationship between ARS-CC scores and several measures of drinking history. Specifically, average self-efficacy to use these drinking-reduction strategies was significantly negatively correlated (all Ps < 0.01) with AUDIT scores [r (352) = − 0.26], S-RAPI scores [r (352) = − 0.18], typical number of drinks per drinking session [r (352) = − 0.25], typical number of days drinking per week [r (352) = − 0.20], perceived difficulty of abstaining in the next 30 days [r (352) = − 0.29] and not feeling in control of one's drinking [r (352) = − 0.32]. Although this lends some criterion validity to the ARS-CC, the modest size of these coefficients indicates that current confidence to employ these strategies accounted for relatively small proportions of variance in these other drinking variables.

To assess discriminant validity of the ARS-CC, we correlated scale scores with measures of non-drinking health-related self-efficacy, alcohol outcome expectancies and sensation seeking. ARS-CC scores were statistically significantly associated with self-efficacy to achieve non-drinking health outcomes [r (352) = 0.21, P < 0.01], but were not significantly correlated with either positive outcome expectancies [r (352) = − 0.07, ns], negative outcome expectancies [r (352) = − 0.10, ns] or sensation seeking [r (352) = 0.00, ns]. We interpret this pattern of weak and generally non-significant associations as providing further support for the discriminant validity of the ARS-CC.

Finally, because social desirability bias might lead participants to report more confidence to employ the listed alcohol reduction strategies, we also evaluated the association of ARS-CC scores with scores on the two subscales of the balanced inventory of desirable responding (BIDR). Self-efficacy was significantly correlated with both self-deception [r (352) = 0.27, P < 0.01] and with impression management [r (352) = 0.20, P < 0.01]. These analyses indicate that one's reported self-efficacy to employ these strategies is associated, albeit weakly, with these two measures of social desirability bias.

Association of gender and binge status with ARS-CC scores

To facilitate comparison with Bonar et al. (2011), we also conducted a 2 (gender) × 3 (binge episode status: 0 episodes/past 2 weeks; one-or-two episodes/past 2 weeks; three-or-more episodes/past 2 weeks) between-subjects analysis of variance using average ratings across the 31 items on the ARS-CC questionnaire as the dependent variable. There was a significant main effect for gender, F(1, 345) = 13.91, P < 0.001, partial η² = 0.04, such that female drinkers (M = 3.74, SD = 0.75) reported higher self-efficacy than male drinkers (M = 3.37, SD = 0.85). There was neither a main effect of episode status, nor was there a significant two-way interaction for gender by episode status on current self-efficacy.

DISCUSSION

As a follow-up to a previous study evaluating the ARS-CC questionnaire (Bonar et al., 2011), we recruited a new sample of binge drinking undergraduates (n = 353) to evaluate the psychometric properties of this self-report questionnaire designed to assess young people's current confidence to employ 31 specific cognitive-behavioral alcohol reduction strategies in a specified drinking situation. Similarly to Bonar et al. (2011), the principal components analysis, internal reliability analysis and mean inter-item correlation of the ARS-CC for this sample suggested that the 31 items comprise a single scale. The significant correlations between ARS-CC scores and several measures of drinking history, and the non-significant correlations of ARS-CC scores with measures of sensation seeking and alcohol outcome expectancies, provide support for criterion and discriminant validity of the measure. These findings are also consistent with the previous study by Bonar et al. (2011) supporting these properties of the ARS-CC.

Although this study supports several elements of reliability and validity of this questionnaire with a new sample, we recruited participants from the same institution as the previous investigation of the ARS-CC. Generalizability to other campuses may be limited to the degree that students’ current confidence to employ specific strategies differs depending on the accessibility of alcohol at the institution one attends, campus-specific alcohol awareness programs and the social acceptability of strategies on different campuses. In addition, depending on the location and campus culture, students may employ other, unlisted strategies to reduce their drinking. Therefore, clinicians and researchers might consider adding an open-ended question to the ARS-CC asking individuals to list personally unique alcohol reduction strategies and to rate their confidence to employ those skills. We also recognize that any self-report questionnaire of this type depends on respondents’ willingness and ability to be insightful and disclosive about their behavior.

Despite these limitations, this second evaluation of the ARS-CC supports its reliability and several aspects of validity. Therefore, both researchers and clinicians might consider using this questionnaire to assess university student drinkers’ self-confidence to employ this wide range of drinking reduction strategies. For example, counselors could have clients complete this instrument prior to an intake interview, and during or in between counseling sessions, to monitor changes in confidence to use these strategies to restrain their heavy drinking. Furthermore, program evaluators could use the ARS-CC as an outcome measure to assess the degree to which educational interventions impact reported confidence to employ specific drinking-reduction strategies in various high-risk drinking situations. Lastly, we encourage additional research to evaluate strategy-specific self-efficacy in non-student and clinical samples.

Funding

We received no external funding to conduct this study.

References