THE STRUCTURE OF DRINKING MOTIVES IN FIRST NATIONS ADOLESCENTS IN NOVA SCOTIA Christopher J. Mushquash, M.A., Sherry H. Stewart, Ph.D., M. Nancy Comeau, Ph.D., and Patrick J. McGrath, Ph.D. OBJECTIVE: The factor structure of the Drinking Motives Questionnaire - Revised (DMQ-R; Cooper, 1994) was examined in a sample of First Nations (i.e., Mi'kmaq) adolescents. RESULTS: Exploratory principal components analysis indicated a three-factor structure (conformity, coping, and positive reinforcement motives), with the positive reinforcement motives of enhancement and social motives not separating into the expected two distinct factors. Moreover, community informants (e.g., school personnel) anecdotally indicated possible wording problems with some of the social motive items for the cultural group. A qualitative methodology - focus group interviews with Mi'kmaq adolescents - was used to explore potential reasons for these observed diff erences in the structure of drinking motives from previous fi ndings in the majority culture (i.e., a measurement problem vs. a real diff erence in the structure of drinking motives in the Mi'kmaq culture). CONCLUSIONS: Qualitative fi ndings support the interpretation that a true social motive for alcohol use does not exist in this cultural/age group and that drinking in social contexts for this group seems less motivated by social affi liation than by enhancement motives (e.g., drinking to party). In Canada, 4.4% of the population is composed of individuals who identify themselves as Aboriginal (Aboriginal people can include Indian, Métis, and Inuit peoples, as recognized in the Constitution of 33 34 VOLUME 15, NUMBER 1 Canada; Statistics Canada, 2001). Unfortunately, the abuse of alcohol and other substances is consistently reported as a major problem in Aboriginal communities (Health Canada, 2003); these communities are well aware of the negative role that alcohol plays in the health of their people. In the Aboriginal Peoples Survey (Statistics Canada, 1991), 73% of First Nations respondents reported that alcohol was a problem in their communities. This situation is not unique to Aboriginal youth in Canada; similar problems are faced by American Indian and Alaska Native (AI/AN) communities in the U.S. For example, Spicer et al. (2003) found that that alcohol dependence for AI men was twice the national average. Data from the Indian Health Service and the Centers for Disease Control and Prevention indicate that alcohol-related hospitalizations among AIs are disproportionately high (Indian Health Service, 1995; Centers for Disease Control and Prevention, 1992). Clearly, there is a need for culturally relevant intervention programming designed to address the issue of alcohol abuse within Aboriginal groups, and for early intervention programming to prevent alcohol problems in Aboriginal young people. When developing interventions for use with Aboriginal groups, the appropriateness of assessment measures derived from majority cultures must be considered. The validity of such measures when used with Aboriginal groups cannot be assumed. Intervention design is dependent upon appropriate assessment; inappropriate assessment may lead to less-than-optimal interventions. For example, when assessing adolescents’ motives for drinking, it is important to appreciate cultural diversity and the eff ects this diversity might have on the validity of psychological measures (Kuntsche, Stewart, & Cooper, 2008). Drinking motives that might be common within a majority culture simply may not apply within other communities or groups. This diff erence may lead to confusion when culturally inappropriate items (designed to tap a specifi c but culturally exclusive construct) on a measure are encountered. If measures are broadly applied across varying cultures, one can reasonably expect that not all items would be relevant for all groups (for more, see Mushquash & Bova, 2007). Adolescents within a group may drink for diff erent reasons, which must be taken into consideration when developing prevention and treatment programs. For example, a treatment approach for an individual who consumes alcohol to cope with negative feelings would be different than that for an individual who consumes alcohol to enhance experiences. Sub-typing drinkers on the basis of their reasons for drinking facilitates the ability to design appropriate and individually DRINKING MOTIVES IN MI'KMAQ ADOLESCENTS 35 specifi c prevention and treatment programs with more accuracy and eff ectiveness (Conrod, Pihl, Stewart, & Dongier, 2000). Such grouping can only be done when considering both the nature of the questions on a measure, and the overall factor structure of the measurement model within the cultural group with which the measure is intended to be used. Cox and Klinger (1988, 1990) proposed a framework for categorizing motives for drinking in which they recognized that people drink to obtain various valued outcomes. This model was adapted by Cooper (1994), who characterized drinking motives along two underlying dimensions. These dimensions refl ect the valence (which can involve positive or negative reinforcement) and the source (internal or external) of outcomes that an individual might hope to achieve by drinking. What emerges is a four-factor model that crosses valence by source, whereby individuals may drink to obtain a positive outcome (positive reinforcement) or to avoid a negative outcome (negative reinforcement), and whereby they may drink to achieve an internal reward (e.g., change in aff ective state) or an external reward (e.g., change in social environment). Each of the four resultant factors represents a distinct motive for drinking (enhancement, social, coping, and conformity). Enhancement motives are internally generated and positively reinforcing. They refl ect the crossing of the positive reinforcement valence, and internal source dimensions (i.e., drinking to enhance pleasurable emotional states). Social motives are externally generated and positively reinforcing. Individuals who are motivated to drink for social reasons are externally controlled, seeking to obtain positive social drinking outcomes (i.e., affi liation with others). Coping motives are internally generated and negatively reinforcing (i.e., drinking to cope with negative emotions), and the remaining motive, conformity, is externally generated and negatively reinforcing (i.e., drinking to reduce social censure; see Cooper, 1994). The ability to identify and classify individuals along these four drinking motives has important implications for intervention and treatment. If the goal of programming is to lessen the harm of drinking, then determining why a person drinks becomes vital. By targeting individuals’ reasons for drinking, the appropriate tools can be provided to enable them to change. Further, some motives have been related to more normative and less risky drinking behavior, while others have been associated with heavier and more problematic drinking, at least in the majority culture. For example, social motives are endorsed more often than any others and are associated with light, infrequent, and 36 VOLUME 15, NUMBER 1 nonprob lematic alcohol use among adolescents and young adults from the majority culture (Cooper, 1994; Stewart et al., in press). Conversely, coping motives have been related to heavier, problematic drinking in these groups (Cooper, 1994). In addition, social motives are related to drinking in social settings while coping motives have been related to drinking alone (Cooper, 1994). Again in contrast to social motives, enhancement motives have been shown to positively predict a pattern of heavy alcohol use and drinking in situations conducive to heavy drinking, and to be related to alcohol problems by virtue of their association with heavier consumption (Cooper; Stewart et al.). Determining why an individual drinks is important to ensure that the right issues are being addressed, whether in educational or therapeutic settings. Study 1 An intervention program (described in detail elsewhere) developed in collaboration with First Nations school-based partners and students, was implemented in two Aboriginal communities in Atlantic Canada (Comeau et al., 2005; Mushquash, Comeau, & Stewart, 2007). This program sought to prevent alcohol misuse by Aboriginal adolescents at risk for alcohol abuse, using a risk-reduction approach. At-risk adolescents were those who scored above one standard deviation on specifi c personality traits (Anxiety Sensitive, Hopelessness/Negative Thinking, and Sensation Seeking) associated with problematic alcohol use as measured by the Substance Use Risk Profile Scale (SURPS; Woicik, Conrod, Stewart, & Pihl, 2008; Krank, Stewart, Wall, Woicik, & Conrod, 2008). As part of the assessment process to screen for eligible intervention participants, the Drinking Motives Questionnaire - Revised (DMQ-R) was administered to students in four high schools (grades 8-12) in two First Nations communities within Nova Scotia. For this study, data collected from Mi’kmaq students were analyzed. Data were collected across 2 school years and pooled to enable an adequate sample size for factor analysis. Method Participants The screening sample consisted of 164 adolescents (84 female, 80 male) from grades 8 to 12. The mean age of the sample was 16.3 years (SD = 1.3) and the average education obtained was grade 10. Of DRINKING MOTIVES IN MI'KMAQ ADOLESCENTS 37 the total pooled sample, 153 students reported using alcohol in the previous 4 months and were included in the analysis (nondrinkers were excluded because the DMQ-R response format requires respondents to be drinkers). Students reported drinking 5-6 or 7-9 drinks per drinking occasion, (M = 3.35, SD = 1.451; on a Likert-type scale of 1 to 5, with each number representing a range of drinks: 1 = 1 to 2 drinks; 2 = 3 to 4 drinks; 3 = 5 to 6 drinks; 4 = 7 to 9 drinks; 5 = 10 or more drinks), and drinking at least 2 to 3 times per month, (M = 2.41, SD = 1.447; on a Likert-type scale of 1 to 5). Measures The DMQ-R is a 20-item self-report measure designed to quantify adolescents’ reasons for drinking alcohol. It is based on an earlier version (DMQ; Cooper, Russell, Skinner, & Windle, 1992) which was designed to measure three distinct drinking motives (coping, enhancement, and social) in adult samples. The revised version was specifi cally designed to measure the four drinking motives described in Cooper’s (1994) model in adolescent samples, and also included the conformity motives subscale. Respondents rate their relative frequency of drinking alcohol for the four reasons (social, enhancement, coping, and conformity), each tapped by fi ve items, on a 5-point scale, with 1 = almost never/never and 5 = almost always/always. For the purposes of this study, if values were missing from participants’ DMQ-R data, the mean of the subscale was used. Each subscale has fi ve values; if two or fewer values were missing from the subscale, the mean of the remaining values was used. If more than two values in a subscale were missing, the participant was excluded from the analysis. Data Analysis In order to explore the factor structure of the DMQ-R, an exploratory principal components factor analytic methodology was employed, because no work had previously been done with the DMQ- R in this population. As well, oblique rotation was used because of the previously observed intercorrelation of the factors on this measure in adolescents (Cooper, 1994) and young adults (Simons, Correia, Carey, & Borsari, 1998; Stewart, Zeitlin, & Samoluk, 1996; Stewart, Watt, Zvolensky, Mushquash, Eifert, & Samoluk, in press) from the majority culture. In the present study, there were mild to moderate correlations (three- factor solution: .260 - .408; four-factor solution: .159 - .415) between the factors. 38 VOLUME 15, NUMBER 1 Results Kaiser’s eigenvalue > 1 criterion for factor extraction supported a four-factor solution (four eigenvalues greater than 1.00; Table 1). When a four-factor solution was examined, 64.77% of the variance was accounted for, but the structure matrix was not interpretable within the DMQ-R framework. Social and enhancement motives loaded on the same factor (1; 41.25% variance explained) and a factor made mostly of coping items emerged (3; 6.98% variance explained). The remaining two factors were composed of items from other motives, with one factor (2; 11.04% variance explained) made mostly of conformity items (and some coping items) and one factor (4; 5.50% variance explained) representing only one item from the measure (see Table 2). Because the fourth factor had only one item with a salient loading, the four-factor solution clearly represented factor over-extraction. Thus, the four-factor solution showed poor simple structure (Thurstone, 1947). Because of the relatively small subject-to-variable ratio, loadings > .60 were considered salient; this criterion is quite strict, but helps ensure the reliability of the solution. Table 1 Eigenvalues for Obliquely-rotated Factor Analysis Eigenvalues Component Total % of Variance Cumulative % 1 8.250 41.251 41.251 2 2.209 11.043 52.294 3 1.396 6.980 59.274 4 1.101 5.503 64.777 When a three-factor solution was examined, slightly less variance was accounted for (59.27%), but the factor solution better refl ected the DMQ-R. As with the four-factor solution, Factor 1 accounted for the most variance and was composed of social and enhancement motive items. Factor 2 was composed of items from the coping motive. Factor 3 was made up of items from the conformity motive (see Table 3). Thus, the three-factor solution was theoretically interpretable and showed excellent simple structure. When a two-factor solution was forced, factors refl ecting positive reinforcement (Factor 1) and negative reinforcement motives (Factor 2) emerged. Although this two-factor solution was more parsimonious, the three-factor solution provided more detail as it separated the negative reinforcement factor into coping and conformity motives. While we did not conduct a confi rmatory factor analysis (which would have allowed us to test if the incremental information provided by splitting coping and DRINKING MOTIVES IN MI'KMAQ ADOLESCENTS 39 conformity motives over a generic negative reinforcement motive was a signifi cant improvement in model fi t), the three-factor solution more closely refl ected the theorized model (Cooper, 1994). A single-factor solution was not examined as it would simply be a refl ection of a general motivation to drink or a proxy measure of drinking frequency. Table 2 Structure Matrix for Obliquely-rotated, Four-factor Solution (N = 153 drinkers) Factor 1 - Factor 2 - Factor 3 - Factor 4 - Enhancement Social Coping Conformity DMQ-R Item Motives Motives Motives Motives Enhancement Motives Subscale 13. Because it gives you a pleasant feeling .834* .347 .438 .172 18. Because it’s fun .876* .252 .323 -.002 7. Because you like the feeling .749* .344 .310 .201 9. Because it’s exciting .779* .399 .239 .187 10. To get high .635* .512 .463 .285 Coping Motives Subscale 17. To forget about your problems .458 .489 .739* .032 1. To forget your worries .181 .285 .694* .012 4. Because it helps you when you feel depressed or nervous .425 .344 .852* .130 6. To cheer up when you are in a bad mood .391 .351 .813* .122 15. Because you feel more self- confi dent and sure of yourself .479 .693* .524 .029 Conformity Motives Subscale 20. So you won’t feel left out .405 .812* .476 .170 12. To fi t in with a group that you like .405 .775* .420 .235 19. To be liked .277 .881* .384 -.030 8. So that others won’t kid you about not drinking .207 .654* .108 .533 2. Because your friends pressure you to drink .103 .156 .101 .844* Social Motives Subscale 5. To be sociable .496 .499 .464 .393 3. Because it helps you enjoy a party .715* .231 .298 .479 14. Because it improves parties and celebrations .780* .275 .382 .064 11. Because it makes social gatherings more fun .841* .391 .423 .112 16. To celebrate a special occasion with friends .766* .245 .329 -.032 Extraction method: Principal Component Analysis Rotation method: Oblimin with Kaiser Normalization *loading >.60 Loadings for Factor 3 have been multiplied by -1.00 to aid in interpretability 40 VOLUME 15, NUMBER 1 Table 3 Structure Matrix for Obliquely-rotated, Three-factor Solution (N = 153 drinkers) Factor 1 - Factor 2 - Factor 3 - Enhancement/ Coping Conformity Social Motives Motives Motives DMQ-R Item Enhancement Motives Subscale 13. Because it gives you a pleasant feeling .837* .419 .278 18. Because it’s fun .866* .327 .139 7. Because you like the feeling .782* .303 .316 9. Because it’s exciting .778* .261 .372 10. To get high .645* .471 .470 Coping Motives Subscale 17. To forget about your problems .463 .758* .279 1. To forget your worries .190 .673* .104 4. Because it helps you when you feel depressed or nervous .440 .795* .173 6. To cheer up when you are in a bad mood .405 .766* .184 15. Because you feel more self-confi dent and sure of yourself .476 .631* .500 Conformity Motives Subscale 20. So you won’t feel left out .409 .597 .679* 12. To fi t in with a group that you like .412 .525 .689* 19. To be liked .268 .581 .669* 8. So that others won’t kid you about not drinking .229 .165 .788* 2. Because your friends pressure you to drink .150 -.042 .499 Social Motives Subscale 5. To be sociable .514 .450 .510 3. Because it helps you enjoy a party .736* .207 .348 14. Because it improves parties and celebrations .777* .374 .178 11. Because it makes social gatherings more fun .839* .431 .291 16. To celebrate a special occasion with friends .757* .340 .119 Extraction Method: Principal Component Analysis Rotation Method: Oblimin with Kaiser Normalization *loading >.60 Discussion The hypothesized four-factor model did not emerge in this analysis. One reason why the solution was diff erent from the expected theoretical model might be sample size. Gorusch (1983) recommended a minimum of fi ve subjects per variable. Higher subject-to-variable ratios are generally better (Tabachnick & Fidell, 2001); small sample sizes yield DRINKING MOTIVES IN MI'KMAQ ADOLESCENTS 41 unstable factors in factor analysis, as correlation coeffi cients estimated from small samples tend to be less reliable (Tabachnick & Fidell). The DMQ-R has 20 items, and only 153 respondents were included in the analysis; this translates to a subject-to-variable ratio of 7.7:1. Given the stringent guidelines with respect to the classifi cation of loadings (salient loadings > .60), this ratio is acceptable. It may be that a three-factor solution better represents the drinking motives of Mi’kmaq youth. In particular, it may be that, within this group, there is an association of drinking in social contexts with enhancement motives leading to a confounding of social and enhancement motives. This would indicate that Cooper’s (1994) model and the DMQ-R need to be modifi ed to ensure that the measure is valid for use with this cultural group. This fi nding speaks to the issue of cultural appropriateness; anecdotal evidence from school personnel indicated potential problems with some of the social motive items. For example, community informants indicated that the word “sociable” (DMQ-R item 5) is not typically used within the Mi’kmaq culture and thus the respondents may not have been able to adequately answer the social motive item that used this term. However, it is unclear whether the fi ndings were related strictly to problems with inappropriate wording and language, or if the three-factor solution was related to a structural diff erence in drinking motives in this sample. Study 2 Because of the emergence of a three-factor model, a qualitative study was conducted to help elucidate the reasons why the social and enhancement motives were not separating into two distinct factors in this cultural/age group. Method Participants This study purposively sampled participants from the screening sample who were identifi ed as being at high risk for alcohol-related problems (i.e., who scored higher than one standard deviation on the SURPS) and participated in the intervention, as well as students who were identifi ed as being at high risk who did not participate in the intervention. In addition, a group of students who participated in the intervention by contributing artwork to the intervention manuals, but who were not in 42 VOLUME 15, NUMBER 1 the high -risk category, participated in the focus groups. This additional group was included to avoid potential confounds associated with sampling only high-risk individuals. For example, it may be that there is no social motive for drinking among high-risk drinkers, and sampling only this group would bias the results in that direction. As well, many of the high-risk students had participated in the intervention and may have been aff ected by the material in such a way as to change their previous motivations for drinking (Mushquash et al., 2007). Apparatus The seven open-ended questions from the Motivational Information – Reasons for Drinking section of the Comprehensive Drinker Profi le (CDP; Marlatt & Miller, 1984) formed the guide for the qualitative interviews. This section of the CDP was previously used in developing the intervention. Specifi cally, it was used to identify scenarios and stories in which community adolescents felt motivated to drink. It was chosen for use in the present study because it captures the source and valence of reasons for drinking, similar to the structure of Cooper’s (1994) model, but in an open-ended manner suitable for use in a group-based interview. In addition to directly querying about reasons for drinking, the CDP asks about the most positive eff ects or consequences associated with drinking, as this is an indirect way to get at why young people drink. In other words, fi nding out what they think are positive eff ects of drinking can help clarify their desired consequences of, or motives for, drinking. Procedure Students were interviewed in small groups of 5 to 10 at their respective schools. A culturally relevant Sharing Circle format, in which students sat quietly and respectfully, taking turns sharing openly without judgement from others, was used to ensure that participants could feel free to communicate their feelings and opinions in a way that was safe; the interviewer was a First Nations young adult from a diff erent group in Canada. All relevant procedures used to protect participants’ confi dentiality were described to the students and, following their assent, the focus group interviews were audiotaped. Upon the recommendation of our community partners, parental consent was not sought so as not to create potential child-protection issues. Audiotapes were later transcribed and data were analyzed for predominant themes. The focus groups could have consisted of students from three pools of potential participants: those who participated in the interventions, those who were eligible but did not participate in the interventions, and those who