Cluster analysis yielded five patterns of respondents. The first was the precontemplation group, who view themselves as not having a problem. The ambivalent group appears to be reluctant or ambivalent about changing their behavior. The participation group seems to be highly invested and involved in change. The uninvolved or discouraged group appears to have given up on the prospects of change and are not involved in attempting to do so. The contemplation group appears to be interested in making changes, are thinking about it, but have not yet begun to take action.

The subtypes were found to differ significantly with respect to the pattern of their alcohol use, the perceived benefits of drinking, and the incidence of negative alcohol-related consequences. The validity of these typologies was largely corroborated in a subsequent cluster analysis of alcohol and drug clients seeking treatment (Carney and Kivlahan 1995).

The second measure, the Stages of Change Readiness and Treatment Eagerness (SOCRATES-Miller, W.R., et al. 1990; Miller and Tonigan 1994), consists of either a 40item version or an abbreviated 20-item version. Like the RTCQ but unlike the URICA, the SOCRATES items are worded specifically in reference to changing drinking behavior. These items are responded to along a 5-point Likert scale (from strong agreement to strong disagreement). The measure has been shown to have adequate levels of internal and testretest reliability as well as construct and criterion validity (Miller and Tonigan 1994).

Conceptually, the SOCRATES assesses the stage of readiness expressed by the individual within the theoretical framework proposed by Prochaska and DiClemente. Factor analytic studies by Miller and colleagues, however, indicated three empirically derived scales: readiness for change, taking steps for change, and contemplation (Miller and Tonigan 1994). Isenhart (1994) similarly found three factors on the SOCRATES, labeled determination, action, and contemplation.

Cluster analyses (Isenhart 1994) also suggests three groups based on the pattern of their factor scores. These were similar in nature to those obtained by DiClemente and Hughes (1990) using the URICA, namely, ambivalent, uninvolved, and active. These groups were found to differ significantly with respect to the pattern and styles of drinking and drinking-related consequences as measured by the Alcohol Use Inventory.

Application of Instruments

The stage of readiness to change has direct implications for the development of initial interventions meant to enhance the likelihood of the client engaging in and complying with treatment. The approach taken by the clinician in attempting this task will differ depending on the client's stage of readiness (Prochaska and DiClemente 1986; Prochaska et al. 1992). As an example, clients who are in the early stages of the behavior change process, in which they are contemplating change and moving toward making a commitment and taking action, will likely benefit most by approaches that increase their information and awareness about themselves and the nature of the problem, lead to self-assessment about how they feel and think about themselves in light of a problem, increase their belief in the ability to change, and reaffirm their commitment to take active steps to change (Horvath 1993; Prochaska et al. 1992).

In addition to being consistent with practice wisdom and theoretical approaches to change, the focus on such awareness-raising factors for those in the precontemplation and contemplation phases is consistent with recent evidence from individuals who had resolved an alcohol problem on their own without the aid of formal treatment. Sobell et al. (1993) found that over half of the recoveries of such individuals could be characterized by a cognitive evaluation of the pros and cons of continued drinking.

For some individuals, the events that led

them to contemplate the need for change or to take steps to seek help may be sufficient for them to stop drinking or modify their alcohol use patterns without more formal treatment (Sobell et al. 1993). For others, brief interventions based on a comprehensive assessment of their addictive behaviors and related life areas, the provision of feedback and advice to the client, and a focus on increasing motivation for change, have been found to increase the likelihood of clients following through on referrals to seek and enter treatment (e.g., Bien et al. 1993; Heather 1989; Miller 1989a).

Alcohol-Related Expectancy Measures and Reasons for Drinking

Clinicians and clinical researchers have been focusing increasingly on the role of cognitive factors in decisions to drink and in drinkers' responses to alcohol (Donovan and Marlatt 1980; Goldman et al. 1987; Oei and Baldwin 1994; Oei and Jones 1986; Wilson 1987a, b; Young and Oei 1993). Two broad categories have implications for the development and maintenance of drinking problems and for potential relapse following treatment: expectations about drinking and expectations about one's ability to cope adequately with problems (self-efficacy expectations).

Alcohol-related expectancies typically refer to the beliefs or cognitive representations held by the individual concerning the anticipated effects or outcomes of consuming alcohol. These expectancies are shaped by past direct or indirect experience with alcohol and drinking behavior (Connors and Maisto 1988a). To the extent that these representations are activated and accessible to the individual in drinking-related situations, they are hypothesized to determine the anticipated outcomes in using alcohol and to mediate subsequent drinking behavior (Rather and Goldman 1994; Stacy et al. 1994).

It is often presumed that individuals drink in order to achieve or enhance the emotional or behavioral outcomes that they expect; thus, these expectancies are viewed as reflective of possible "reasons for drinking." Individuals differ with respect to both their experiences with alcohol and drinking and their resultant beliefs and expectations about alcohol's effects.

When expectancies serve a functional role in maintaining problematic drinking behavior, clients may be assigned to treatment strategies designed to challenge or modify their beliefs about alcohol's effects and to substitute more adaptive or realistic expectations. Decreases in positive expectancies associated with alcohol would be expected to decrease drinking behavior (Connors and Maisto 1988a; Connors et al. 1992, 1993; Darkes and Goldman 1993; Oei and Baldwin 1994; Oei and Jones 1986).

Instruments

A number of measures of alcohol-related beliefs and expectancies have been developed and are available to help the clinician determine the nature, strength, and valence of these beliefs. The Alcohol Expectancy Questionnaire* (AEQ-Brown et al. 1987a; Brown et al. 1980) is the most widely used alcohol expectancy measure in both research and clinical settings.

The AEQ is a 90-item self-report form, presented with a forced choice (e.g., agreedisagree) response format, that assesses a diverse array of anticipated experiences associated with alcohol use. It was developed empirically by refining a pool of verbatim statements of adult men and women aged 15-60 years, with diverse ethnic backgrounds and drinking histories (from nondrinkers to chronic alcoholics). The adult version is designed to assess the domain of alcohol reinforcement expectancies and consists of six factor-analytically derived subscales: positive global changes in experience, sexual enhancement, social and physical pleasure,

assertiveness, relaxation/tension reduction,

and arousal/interpersonal power.

The AEQ has been evaluated in clinical and nonclinical populations and has a high level of internal consistency and test-retest reliability. In addition to concurrent validity, the total-score and subscale scores have differentiated alcoholic from nonalcoholic respondents and predicted current and future drinking practices, persistence and participation in treatment, and relapse following treatment (Brown 1985a, b; Brown et al. 1987a).

Despite the systematization brought to the assessment of alcohol expectancies by the AEQ, investigators and clinicians have noted a number of theoretical and practical limitations in its utility. These include its reliance on a forced-choice response format that does not reflect the strength of the expectancies; a confounding of global or general beliefs with personal ones; its focus on positive outcome expectancies without assessing anticipated negative outcomes; its restriction to a single "dose" or level of alcohol in the instructions (e.g., a "few drinks"), thus precluding examination of variation in expectancies over different dose levels; and the lack of any measure of frequency of occurrence or personal importance associated with each of the expectancies (e.g., Adams and McNeil 1991; Collins et al. 1990; Connors et al. 1992; Leigh 1989b; Leigh and Stacy 1991, 1993; Oei et al. 1990; Southwick et al. 1981). These concerns have led to the development of several subsequent expectancy measures that attempt to address one or more of these limitations.

The Alcohol Effects Questionnaire* (AEFQ-Rohsenow 1983), a revision and extension of the AEQ, was developed to meet several of these concerns. It is briefer (40 true/false items rather than 90); it assesses undesirable effects of alcohol (impairment and irresponsibility) as well as positive reinforcing effects; and it assesses only personal beliefs (beliefs about the effects of alcohol on oneself) rather than mixing personal beliefs

with general beliefs (beliefs about the effects of alcohol on people in general).

The AEFQ was constructed by taking the five items that loaded most highly on the six factors of the AEQ, adding two items assessing verbal aggression and deleting from the aggression scale one item that had loaded on two factors, and adding five items assessing cognitive and physical impairment and four items assessing carelessness or lack of concern about consequences. All items were then reworded to reflect personal beliefs.

The AEFQ consists of eight rational scales: global positive, social and physical pleasure, sexual enhancement, power and aggression, social expressiveness, relaxation and tension reduction, cognitive and physical impairment, and careless unconcern. Internal consistency indices across subscales ranged from 0.49 to 0.74 for college student drinkers and from 0.37 to 0.85 among alcoholics in treatment.

Factor analysis of the AEFQ on college students (Rohsenow 1983) largely supported the first six rationally derived factors and combined the two negative scales into one factor. The AEFQ has been used largely as a research instrument to explain or predict other responses of the sample, such as aggression after drinking (Rohsenow and Bachorowski 1984) and cue reactivity (Rohsenow et al. 1992).

Recently, George and colleagues (1995) modified and extended the AEFQ into the AEQ-3. (i.e., third revision of the Alcohol Expectancy Questionnaire). The objective was to maintain the benefits of the AEFQ (e.g., brevity and negative expectancies) while shifting the response format to a 6point rating scale (from strongly agree to strongly disagree) to allow information about strength of endorsement. The structure derived from confirmatory factor analysis of the AEQ-3 was found to be relatively consistent with that proposed by Rohsenow

(1983) and was relatively invariant across gender and ethnic groups.

Another measure of expectancies is the Drinking Expectancy Questionnaire* (DEQYoung and Knight 1989; Young et al. 1991a). It also attempts to improve on the AEQ by phrasing items consistently in the first person, by measuring both positive and negative expectancies, and by balancing the valence of items selected for the questionnaire by providing a multiple response format (Young and Knight 1989).

The DEQ consists of 43 items developed using both community and clinical populations. Each item is rated on a 5-point scale from strongly disagree to strongly agree. Five subscales, derived from factor analysis, relate to specific alcohol expectancies of assertion, affective change, sexual enhance ment, cognitive change, and tension reduction. A sixth factor, dependence, is more general and relates to perceived level of alcohol involvement. The total score and the subscale scores of the DEQ have been found to correlate with measures of frequency of drinking, but not quantity consumed, in a community sample (Lee and Oei 1993a). As an example, those who expected greater negative affective states when drinking drank both their usual and maximum amounts of alcohol less often.

The Alcohol Beliefs Scale* (ABS-Connors et al. 1987, 1992; Connors and Maisto 1988b) is a two-part, 48-item questionnaire. It attempts to incorporate information concerning strength of endorsement, doserelated changes in the anticipated effects of alcohol, and the perceived utility of alcohol in inducing a number of emotions or behaviors. On part A of the scale (26 items), subjects indicate the extent to which each of three different amounts of alcohol (one to three standard drinks, four to six standard drinks, and "when drunk") increase or decrease behaviors and feelings such as judgment, problem solving, depression, aggression, stress, and group interaction. The ratings are

made on an 11-point scale ranging from a “strong decrease in behavior or feeling" to a "strong increase in behavior or feeling"; a rating of zero is used to indicate no change in the behavior or feeling as a result of drinking. Four domains have been derived from the items contained in part A: control issues, sensations, capability issues, and social issues.

On part B of the scale (22 items), drinkers rate how useful the consumption of each of the three doses of alcohol would be for a variety of reasons (e.g., to relax, to become more popular, to become uninhibited, to relieve depression, and to forget worries). These estimates are also made on an 11-point scale ranging from "not at all useful" to "very useful." The factors derived from part B have been labeled as useful in feeling better, useful for being in charge, and useful for alleviating aversive states.

Results suggest that alcoholics differ from problem drinkers and nonproblem drinkers with respect to the expected effects of alcohol and its anticipated utility. In general, alcoholics anticipated less impairment on the control and capability factors. A dose-response relationship was noted, with all drinkers expecting increased impairment with increasing doses. An interaction between drinker group and dose was found on a number of subscales of part B, suggesting differences in the perceived utility to induce moods and behaviors as a function of severity of drinking problem and amount consumed. As an example, higher doses of alcohol were perceived as increasingly useful in reducing emotional distress, with the magnitude of the increases being greatest for alcoholics. There also appears to be an interaction with respect to perceived effects and utility across doses as a function of gender and ethnicity (Connors et al. 1988).

Fromme, Stroot, and Kaplan (1993) recently developed the Comprehensive Effects of Alcohol (CEOA) scale through exploratory factor analysis. This process identified 22