In response to the need for research that incorporates multiple aspects of theory into a testable framework, this study attempted to replicate and extend the results of Cooper, Russell, Skinner, Frone, and Mudar (1992). A modified stressor vulnerability model of stress-related drinking was tested in a homogeneous sample of 65 male and female undergraduate student drinkers.
Total weekly consumption of alcohol was used as the criterion measure, whereas family history of alcoholism (Adapted SMAST: Sher & Descutner, 1986), alcohol outcome expectancies/valences (CEOA: Fromme, Stroot, & Kaplan, 1993), perceived stress (PSS: Cohen, Kamarck, & Mermelstein, 1983), and coping dispositions (COPE: Carver, Scheier, & Weintraub, 1989) were used as the predictor variables. The proposed modified model postulates that expectancies play a proximal mediating role in stress-related drinking, whereas gender, family history of alcoholism, and coping all play a distal moderating role.
Hierarchical multiple regression procedures were then performed to evaluate the model. The results failed to support the hypothesized model. Specifically, expectancies emerged as a distal rather than proximal predictor of stress-related drinking, and family history of alcoholism did not moderate stress-related drinking. In contrast, gender and coping styles emerged as the most powerful predictors in the model. Despite the shortcomings of the proposed model, the present results offer an alternative interpretation as to what constitutes the stressor vulnerability model of stress-related drinking.
Stress as a Causal Factor in Drinking One of the common stereotypes about the effects of alcohol involves the drug’s capacity to act as a stress antagonist. Conger (1956) has proposed a theory, known as the tension reduction hypothesis (TRH) of drinking, to support this notion. Essentially the theory holds that alcohol’s sedative action on the central nervous system serves to reduce tension, and because tension reduction is reinforcing, people drink to escape it (Marlatt & Rohsenow, 1980).
Strong evidence to support the validity of the theory comes from epidemiological findings which indicate that the prevalence of anxiety disorders in alcoholics ranges from 16 to 37%, compared to a rate of only 4-5% in the general population (Welte, 1985). Notwithstanding, there seems to be a subset of people for whom the predictions of the TRH do not hold. For instance, in a study conducted by Conway, Vickers, Ward, and Rahe (1981) it was found that the consumption of alcohol among Navy officers during periods of high job demands was actually lower than the consumption during low-demand periods. In addition, other studies (i. , Mayfield, 1968; Mendlson, Ladou, & Soloman, 1964) have shown that some drinkers actually consider alcohol as a tension generator rather than a tension reducer.
Overall, when taking into account these conflicting findings, it seems prudent to find some middle ground. The solution to this problem than is a modified version of the TRH, specifying the conditions under which stress will lead to an increase in drinking. Moderating and Mediating Factors in Stress Induced Drinking. In addition to stress, several other variables have been shown to be crucial in determining an individual’s drinking behavior.
These variables include gender of drinker (gender), coping behavior of drinker (coping), and alcohol outcome expectancies of drinker (expectancies). In the following discussion, the importance of each of these variables to drinking will be considered first, followed by an evaluation of these as potential moderators or mediators of stress in drinking. Differential Gender Drinking Behavior It has been repeatedly demonstrated that significant differences exist between the drinking patterns of men and women (Hilton, 1988).
In a comprehensive survey of US drinking habits conducted by the US National Center for Health Statistics in 1988, significant gender differences were found in three areas as pointed out by Dawson and Archer (1992). The first significant difference pertained to the number of male and female current drinkers. Roughly 64% of all men were current drinkers in comparison to 41% of all women. The second and third significant differences concerned the quantity of alcohol consumed. Men were more likely to (a) consume alcohol on a daily basis and (b) be classified as heavy drinkers. Men’s daily average of ethanol intake (17. rams per day) was almost twice as high as women’s (8. 9 grams per day). Even when an adjustment for body weight was made (females require less ethanol than males to achieve a similar increase in blood alcohol level), men’s consumption was still 53% greater than women’s.
With regards to drinking classification, males were classified substantially more often than females as heavy drinkers (i. e. the number of males who drank five or more drinks a day was 88% greater than the corresponding number of females). Furthermore, as the classification measures became stricter so did the disparity between male and female heavy drinkers increase (i. , the ratio of male to female heavy drinkers increased by a factor of 3 as the definition of heavy drinker was changed from five drinks or more a day to nine drinks or more a day). Gender as a Moderating Factor of Stress To understand why men and women drink differently requires an understanding of the prevailing socialization practices (Dohrenwend & Dohrenwend, 1976; Horwitz & White, 1987). According to this sociological view, “women have been socialized to internalize distress, whereas men have been socialized to externalize distress” (Cooper, Russell, Skinner, Frone, & Mudar, 1992; P. 140).
Therefore, women tend to cope with stress by utilizing personal (internal) devices such as emotion, rather than impersonal (external) devices such as alcohol, which are used more often by men. In addition, men and women also hold differential expectations about the effects of drinking. Several studies (i. e. Abrams & Wilson, 1979; Sutker, Allain, Brantly, & Randall, 1982; Wilson & Abrams, 1977) have shown that “although pharmacological effects appear to be similarly stress reducing for both sexes, the belief that alcohol has been consumed may actually increase distress among women” (Cooper et al. 992; P. 140). Therefore, it seems plausible that females actually expect to experience some form of distress from drinking as opposed to males’ expectation to experience tension reduction from drinking (Rohsenow, 1983). Differential Coping Styles in Drinking Considerable evidence has been accumulated in support of the notion that certain methods of coping are more likely to be associated with problem drinking than others (Moos, Finney, & Chan, 1981).
This has led to the development of social learning theory which postulates that abusive drinkers differ from relatively healthy drinkers in (a) their capacity to effectively cope with stressors and (b) in their beliefs about drinking (Abrams & Niaura, 1987). In general, two types of coping responses have been shown to predominate in most situations (Folkman & Lazarus, 1980). The first type, problem-focused coping (also known as approach coping), is directed at either solving the presenting problem or altering the source of the stress (Carver, Scheier, & Weintraub 1989).
The second type, emotion-focused coping (also known as avoidance coping ), attempts to reduce the unpleasant emotional feelings which accompany the stressor (Carver et al. , 1989). Even though people usually use both methods in response to a given stressor, the former type will tend to predominate when people feel that the situation is changeable, whereas the latter type will tend to predominate when people appraise the situation as unchangeable (Folkman & Lazarus, 1980).
People who predominantly resort to avoidance coping have been shown to display pathological drinking behavior much more than those who utilize approach coping (Cooper, Russell, & George, 1988; Cooper et al. , 1992). Those who typically resort to avoidance coping (a group which consists of up to 25% of all drinkers), report that they do so in order to regulate negative emotions (Cahalan, Cisin, & Crossley, 1969; Mullford & Miller, 1963; Polich & Orvis, 1979). The strongest evidence to support this contention comes from studies which have investigated post-treatment relapse in alcoholics.
In three such studies (Marlatt, & Gordon, 1979; Moos et al. , 1981; Moos, Finney, & Gamble, 1982), it was found that individuals were more likely to relapse in situations which elicited unpleasant emotional states. Coping as a Moderating Factor of Stress The key to understanding the differential impact of avoidance and approach coping on drinking lies in the availability of an effective coping response to a given stressor (Cooper et al. , 1992). By definition, people who utilize approach coping mechanisms to deal with their stress, engage in concrete problem solving which serves to actively reduce the amount of stress.
By contrast, people who rely on avoidance coping may manage to reduce their distress, but they tend to do so by distracting themselves from the stress. Therefore, it is not surprising that drinking should appeal more to those who predominately use avoidance coping, because the consumption of alcohol serves as a substitute action which can distract from the stress. When viewed from a social learning perspective (Abrams & Niaura, 1987), it can be seen that “alcohol use serves as a general coping mechanism invoked when other presumably more effective coping responses are navailable” (Cooper et al. 1992; P. 140). Evidence to support this idea comes from studies (i. e. , Higgins & Marlatt, 1975; Hull & young 1983; Marlatt, Kosturn, & Lang, 1975) which have investigated drinking in response to negative affects, when no coping alternative was present. For instance, Marlatt et al. (1975) have shown that drinkers who were provoked and were unable to retaliate drank significantly more at a subsequent taste rating task than drinkers who had the option to retaliate (Cooper et al. , 1988). Differential Expectancies About Drinking
Alcohol outcome expectancies (AOE) can be thought of as the beliefs people hold about the effects of drinking (Goldman, Brown, & Christiansen, 1987). These expectancies first develop in childhood as indirect learning experiences (e. g. , media, family modeling, peer influence ) and, as a result of increased direct experiences with the pharmacological effects of alcohol, become more refined (Christiansen, Goldman, & Inn, 1982; Christiansen & Goldman, 1983; Christiansen, Goldman, & Brown 1985; and Miller, Smith, & Goldman, 1990).
The expectancies that people hold about alcohol have been shown to predict alcohol consumption in a variety of settings (Goldman, Brown, & Christiansen 1987). Brown, Goldman, Inn, and Anderson (1980) have shown that light drinkers typically hold global expectancies about alcohol (i. e. alcohol affects multiple factors), whereas heavy drinkers typically hold more specific expectancies, such as alcohol’s ability to increase sexual and aggressive behavior.
Furthermore, Brown (1985a) has shown that people who hold the expectancy that alcohol enhances social experience are less likely to be problem drinkers than people who drink with the expectancy of tension reduction. It is important to note, however, that AOE may “vary with learning context, personal characteristics of the drinker, amount of alcohol consumed, and other addiction risk factors” (Brown, 1993; P. 58). Expectancies as a Mediating Factor of Stress Although it is well established that AOE differentially predict drinking behavior (Brown, 1993), very little is known about how they exert their effects.
To date, most of the research suggests that AOE (gender specific) directly predict alcohol consumption and, as such, are thought to play a mediational role (Brown, 1993). Intuitively, it makes sense that people who hold the expectancy that alcohol can alleviate their stress should drink more than people who do not hold this expectancy. However, little research has been conducted thus far to support this contention. Prior to Cooper et al. (1992), only one study (McKirnan & Peterson, 1988) investigated the role of expectancies in stress-induced drinking.
The study tested a stress-vulnerability model among homosexual men, who show culturally specific stressors and vulnerability (i. e. , homophobic iscrimination). It was found that tension reduction expectancies significantly predicted drinking among individuals who experienced “negative affectivity” stress (i. e. , low self-esteem). Although the Mckirnan and Peterson (1988) study found that expectancies exacerbated stress, the utility of the findings is limited because of the use of a non-representative sample of gay males, and non-standard measures of stress (Cooper et al. , 1992).
The Synthesis of Gender, Coping & Expectancies in Stress-Related Drinking As was discussed previously, gender, coping, and expectancies are thought to play a significant role in stress-related drinking. Nevertheless, the bulk of literature in this area has typically investigated these factors in isolation from each other (at best, only two of these factors have been combined simultaneously). Since stress-related drinking , however, is such a complex phenomenon (recall that the tension reduction hypothesis of drinking does not apply universally) it is necessary to integrate these factors in order to gain a complete, holistic picture.
The only study which has combined all three factors simultaneously was the landmark study of Cooper et al. (1992). The study tested an interactional model of stress-related drinking which postulated that “exposure to environmental stressors is most strongly related to alcohol use and abuse among vulnerable individuals”, such that, “Vulnerable individuals are more likely to be male, to hold strong positive expectancies for alcohol’s effects, and to have limited adaptive coping responses” (Cooper et al. , 1992; P. 141).
The results supported a stressor vulnerability model of drinking. As expected, it was found that men were more likely to drink than women by virtue of their gender role socialization. More importantly, however, it was also found that, for problem drinking to occur in men, a second vulnerability factor must often be present. In particular, men who either held strong positive expectancies or relied on avoidant forms of coping were more likely to be problem drinkers than men who did not possess these attributes. (Cooper et al. , 1992).
With respect to expectancies, it was shown again that both men and women who held strong positive AOE, drank significantly more then men and women who did not. Much more important, however, was the finding that “expectancies appeared to function as stressor vulnerability factor among men but not among women” (Cooper et al. , 1992; P. 148). Finally, with regards to coping, it was confirmed that coping styles play an important role in problem drinking. However, significant interactions with gender and expectancies were also indicated.
Men who relied on avoidant forms of coping were more likely than women to be vulnerable to stress induced drinking. Similarly, stressors were much more likely to elicit problem drinking among individuals who were both high in avoidance coping and positive AOE, than amongst individuals who were only high in avoidance coping. (Cooper et al. , 1992). The purpose of the present study is to extend and modify the work of Cooper et al. (1992) in an attempt to clarify the role of stress in alcohol consumption, with respect to the interactional stressor vulnerability model of drinking.
Specifically, the Cooper et al. (1992) study was limited to the investigation of gender, coping, and expectancies in stress-related drinking. Given that Family history of alcoholism (FH) has been shown to play a significant role in drinking (i. e. , Cotton, 1979; Goodwin, 1988; Hill, Nord, & Blow 1992; Ohannessian & Hesselbrock, 1993), it is appealing to investigate the role of FH as an additional vulnerability factor [as suggested by Cooper et al. (1992)]. Moreover, the Cooper et al. (1992) study conceptualizes gender, coping, and expectancies as moderators of stress-related drinking.
Given that expectancies directly predict alcohol consumption (as discussed previously), a modified interactional model is proposed such that gender, coping, and family history play an indirect moderational role in predicting stress-related drinking; whereas expectancies play a direct mediational role as conceptualized by Figure 1 below. All subjects in this study were undergraduate psychology students from a large Canadian university. The initial sample consisted of 84 volunteers. For the purpose of this study, only those subjects who drank at least once a weak were included.
A total of 65 out of 84 subjects (77. 4%), aged 19 years and over, successfully met this criterion. The sample consisted of a roughly equal number of 31males (47. 7%) and 34 females (52. 3%), who were predominantly Caucasian (64. 6%). More than three-quarters (75. 4%) of the subjects were in their first year of studies, and were mostly employed part-time (60. 0%). Nearly three-quarters (72. 3%) of those who were employed received an annual income smaller than $ 10,000. The mean age at which subjects first consumed alcohol was 14. 7, whereas the mean age at which they began to drink regularly was 17. Subjects total weekly consumption of alcohol averaged 11. 1 drinks.
Measures used in the present study were embedded in a general assessment battery that was a part of a larger research project. For the purpose of the present study, the following measures, administered in a fixed order, were employed to assess the variables of interest. Weekly Alcohol Consumption. Subjects were given a chart which contained the days of the week. For each day they were instructed to indicate the number of standard alcoholic drinks and the amount of time it would take to consume these drinks in a typical week.
A standard alcohol drink was defined as either a regular size can/bottle of beer, 1. 5 ounce shot of liquor, or a 5 ounce glass of wine. Subjects who drank less than once a month were instructed to skip this section. The total number of drinks in one week was summed and used as the dependent variable. Adapted Short Michigan Alcoholism Screening Test (Adapted SMAST) The adapted SMAST (Sher &Descutner, 1986) is a 13 item self-report questionnaire designed to measure family history of alcoholism.
Specifically, the questionnaire assesses the extent of an individual’s mother’s and father’s alcohol abuse. Assessment is based on a two point scale consisting of 0=no and 1=yes. For the purpose of the present study only 10 items were used, and the mother/father answer categories were extended to biological mother/father and step or adoptive mother/father. Comprehensive Effects of Alcohol (CEOA) The CEOA (Fromme, Stroot, & Kaplan, 1993) is a 38 item self-report questionnaire designed to assess alcohol outcome expectancies and their subjective valence.
It is composed of seven expectancy scales, four positive (sociability, tension-reduction, liquid-courage, and sexuality) and three negative (cognitive-behavioral impairment, risk and aggression, and self perception). Expectancy assessment is based on a four point scale from 1=disagree to 4=agree. The valence of these expectancies is assessed on a five point scale from 1=bad to 5=good. Both items and instructions were carefully worded to ensure that the elicited expectancies were neither dose-specific, nor situation specific.
Perceived Stress Scale (PSS). The PSS (Cohen, Kamarck, & Mermelstein, 1983) is a 14 item self-report questionnaire designed to assess the degree to which situations in one’s life are appraised as stressful. An equal number of 7 positive and 7 negative statements make up the questionnaire. Assessment is based on a five point scale from 0=never to 4=very often. Scores are obtained by reversing the scores on the seven positive items (i. e. , 0=4, 1=3, 2=2, etc. ), and then summing across all 14 items. The COPE (Carver et al. 1989) is a 53 item self-report questionnaire designed to assess ndividual coping dispositions. The questionnaire is comprised of 14 scales which are categorized into three coping styles: Problem-Focused Coping (Active coping, Planning, Suppression of competing activities, Seeking social support for instrumental reasons, and Restraint coping), Emotion-Focused Coping (Acceptance, Seeking social support for emotional reasons, Positive reinterpretation, Turning to religion, and Focus on and venting of emotion), and Less than Useful Coping (Denial, Behavioral Disengagement, and Mental Disengagement).
For the purpose of the present study the Alcohol-drug disengagement scale was excluded from these categories, and was treated as a separate category called Drinking to Cope. Assessment is based on a four point scale from1=I usually don’t do this at all to 4=I usually do this a lot. Both items and instructions were worded such that dispositional , rather than situational, styles of coping were assessed. All participants were recruited from undergraduate psychology courses at York University. The questionnaire was administered in a classroom setting.
Participants completed the questionnaire in a group format of mixed sex ranging in size from 10 to 30 individuals. Informed consent was obtained from all participants, and a phone number was provided in case any concerns arose. The complete questionnaire required approximately 40 minutes to administer. Respondents were compensated for their time by being entered in a lottery with a 1 in 50 chance of winning $ 50. 00. Table 1 presents zero-order correlations, computed for all relevant study variables.
Conceptually variables may be grouped into one of five categories: weekly drinking (variable 1), perceived stress (variable 2), family history of alcoholism (variable 3), coping variables (Variables 4-7), and expectancy variables (variables 8-21). Examining the pattern of correlations between these variables suggests several conclusions. First, family history of alcoholism was neither significantly correlated with perceived stress nor with weekly drinking, suggesting that family history of alcoholism is not important in stress-induced drinking.
Second, several coping variables were significantly correlated with either weekly drinking and/or perceived stress. Specifically, drinking to cope was significantly positively correlated with both weekly drinking (r = . 420) and perceived stress (r = . 310), less useful coping was significantly positively correlated (r = . 674) with stress, and problem focused coping was significantly negatively correlated (r = -. 327) with weekly drinking. These findings suggest that coping variables play an important role in stress-related drinking.
Finally, only one expectancy variable, the valence expectancy for cognitive and behavioral impairment, was significantly correlated (r = . 340) with weekly drinking, but not with perceived stress. However, several expectancy variables were significantly positively correlated (. 357 * r * . 517) with drinking to cope. These findings suggest that expectancies are more likely be a distal, rather than a proximal predictor of stress- related drinking. Hierarchical multiple regression analyses were employed to test the model depicted in Figure 1.
Table 2 contains summary statistics for the stepwise regression used to identify the predictor variables of weekly drinking. As can be seen from Table 2, gender emerged as the most important predictor variable accounting for over 28% of the variance. The coping variables of drinking to cope and problem-focused drinking were also significant, and accounted for an additional 12% and 8% of the variance, respectively. Further multiple regression analyses were used to determine which variables predicted drinking to cope, and problem-focused coping, respectively.
Table 3 shows that the expectancy for risk accounted for over 26% of the variance in predicting drinking to cope, with the expectancy for tension and perceived stress accounting for an additional 16%. Table 4 shows that emotion-focused coping accounted for over 34% of the variance in predicting problem-focused drinking, with the expectancy valence for self perception accounting for an additional 8%. Figure 2 summarizes the direct effects estimated in the foregoing series of multiple regression analyses. The primary purpose of the present study was to replicate and extend the results of Cooper et al. 992). Specifically, the Cooper et al. (1992) stressor vulnerability model was modified to include family history of alcoholism as an additional moderating variable, and expectancies were conceptualized as a mediating rather than a moderating variable.
The model was then tested in a college population sample. Overall, the results of this study fail to support the proposed model, suggesting that expectancies play a distal rather than proximal role in stress-related drinking. Furthermore, contrary to previous findings (i. e. , Cotton, 1979; Goodwin, 1988; Hill et al. 992; & Ohannessian & Hesselbrock, 1993), family history of alcoholism did not play a significant role, either directly or indirectly, in predicting drinking. In contrast, the present results indicate that gender and coping are the most powerful explanatory variables in the model. With respect to coping, only drinking to cope and problem-focused coping directly predicted weekly drinking. Expectancies also emerged as an important explanatory variable, however, their influence was exerted via indirect pathways (i. e. , via drinking to cope and problem-focused drinking).
Likewise, perceived stress and emotion-focused coping also emerged as peripheral predictors (via similar pathways) of weekly drinking. Although the present results fail to support the proposed model, they are nevertheless fairly consistent with the stressor vulnerability model proposed by Cooper et al. (1992). Specifically, in keeping with Cooper et al. ‘s (1992) results, the present findings indicate that stress does not directly induce drinking, and that another vulnerability factor must coexist with stress for drinking to occur.
However, there is some apparent discrepancy with respect to the specifics of these additional vulnerability factors, particularly concerning the subtypes of coping mechanisms which affect stress-related drinking. Whereas Cooper et al. (1992) found that individuals who were high in avoidance coping tended to drink more in response to stress, the present study found that drinking to cope was the primary coping mechanism which moderated stress-related drinking. In addition, the present study also found that the moderating influence of expectancies was not limited to positive expectancies, as suggested by Cooper et al. 992). Instead, the negative expectancy for risk and aggression appears to act in concert with the positive expectancy of tension reduction to moderate stress-related drinking.
It is important to note that many of the discrepancies between the present findings and the findings of Cooper et al. (1992) are possibly due to significant differences in the measurement instruments employed. First, with regards to the dependent measure, alcohol consumption, the criterion used to define one standard drink in the Cooper et al. 992) study required significantly smaller amounts of alcohol than did the present study (i. e. , 1 oz. versus 1. 5 oz. of hard liquor, or 4 oz. versus 5 oz. of wine). Therefore, it is possible that by utilizing stricter alcohol consumption criteria in the present study, certain findings that were found to be significant in the Cooper et al. (1992) study were not significant here. Second, with regards to the measurement of expectancies, the Cooper et al. (1992) study measured only positive expectancies.
By contrast, the present study measured both positive and negative expectancies, and it was found that both are important in stress-related drinking. Third, with respect to coping, the present study elaborated on the three coping styles used in Cooper et al. (1992) by adding a fourth category, drinking to cope. This coping style proved to be a superior predictor of stress-related drinking than any of the other coping dispositions. Finally, with respect to the measurement of stress, the Cooper et al. (1992) study used a life events scale to measure stress.
This measure of stress is inferior to the perceived stress scale, because it only measures the objective occurrence of events, and not the subjective interpretation of whether the individual perceives these events to be stressful (Cohen et al. , 1983). As such, the discrepancy between the overall results obtained in the present study and that reported by Cooper et al. (1992) may be due, in part, to the fact that differing constructs were investigated. Two limitations of the present study should be noted.
The first and possibly most significant limitation is the relatively small sample size of the study. This limitation precluded more in-depth statistical analysis, particularly with respect to analyzing interactions. For instance, even though Figure 2 provides the impression that drinking to cope plays a mediating role in stress-related drinking, this conclusion remains invalid without the possibility of analyzing how drinking to cope interacts with gender and problem-focused coping. Second, the present study examined a homogeneous college sample with respect to age, social class, education, and race.
This limits the generalizability of the findings. Several issues are raised by the present findings that could be addressed in future research. Perhaps most importantly, there is a need to determine which instruments provide the most accurate measurement of the constructs in question. This would create a standard set of instruments that could be universally applied in future investigations. Further investigation of negative expectancies would also be useful given the import of the expectancy for risk and aggression in the present data.
Similarly, the linkages among the determinants in this model could be further clarified by examining the potential mediating role of drinking to cope. Ultimately, to elucidate the nature of stress-related drinking, other potential moderators (i. e. , income, race, religion ) must also be investigated to provide a more stringent test of the stressor vulnerability model. In conclusion, these results indicate that the stressor vulnerability model of stress-related drinking as suggested by Cooper et al. (1992) is somewhat imprecise. A more refined model must pay closer attentionto the influence of drinking to cope and negative expectancies.