were related to frequency of drinking but not to quantity of alcohol consumed. Those individuals who expected greater negative affective states while drinking drank their usual and maximum amounts less often, while those who had higher expectations of poor control over drinking drank their usual and maximum amounts more often. The complexity of these relationships, as well as similar ones found in a college sample (Baldwin et al. 1993), likely reflect the nature of the interactions between self-efficacy and alcohol expectancies and their influence on drinking behavior. This area clearly warrants further investigation.

Perceived Locus of Control of

Drinking Behavior

A final set of factors that have played a role in some cognitive-behavioral models of problem drinking and alcoholism is the individual's perception of control (e.g., Donovan and O'Leary 1983; Carlisle 1991). The concept of locus of control, originally developed by Rotter (1966, 1975), refers to whether an individual believes that important life events are under personal control (internal locus of control) or under the influence of chance, fate, or powerful others (external locus of control). Rotter suggested that the predictive utility of this construct is increased by using measures directly related to the behavior under consideration rather than assessing a more generalized perception of control.

To this end, Keyson and Janda (unpublished) developed a locus of control scale related to drinking behavior. The DrinkingRelated Locus of Control Scale* (DRIE-published in Lettieri et al. 1985) assesses the individual's perceptions of control with respect to alcohol, drinking behavior, and recovery. It is a 25-item self-report questionnaire adapted from Rotters conceptual model and assessment method.

In a forced-choice format, individuals are

asked which of two responses best matches their beliefs. These response options include an internal (“I have control over my drinking") and an external ("I feel completely helpless when it comes to resisting a drink”) alternative. The scale is scored in the direction of increasing externality.

Donovan and O'Leary (1978) found that the DRIE has a high degree of reliability; is multidimensional, having empirically defined factors assessing perceived control over interpersonal factors, intrapersonal factors, and general factors associated with drinking; and differentiates between alcohol-dependent individuals (more external scores) and nondependent drinkers. An external locus of control was associated with more physical, social, and psychological impairment from drinking.

The scale has been found to differentiate between groups with varying histories of drinking problems and sobriety or with varying degrees of commitment to change, with more internal scores being associated with longer periods of sobriety or more advanced commitment in the recovery process (Mariano et al. 1989; Strom and Barone 1993). Consistent with this pattern, the perception of control appears to become more internal over the course of alcohol treatment; individuals with more external perceptions are also more likely to drop out of treatment prematurely (Jones 1985; Prasadarao and Mishra 1992).

There appears to be a complex interactive relationship between the primary reasons alcoholics give for their pretreatment drinking and their drinking-related locus of control in predicting posttreatment relapse (Kivlahan et al. 1983), suggesting possible avenues of treatment matching within a relapse prevention framework. Following treatment, alcoholics with an internal drinking-related locus of control are less likely to relapse, drink less and have a shorter drinking episode if they do relapse, and have a better overall drinking-related outcome than

alcoholics with an external DRIE score (Koski-Jannes 1994).

The DRIE represents an additional measure to consider in the assessment of cognitions related to the maintenance of, cessation of, and relapse to drinking behavior. Its relationship with the other cognitive constructs discussed, namely alcohol-related outcome expectancies and self-efficacy expectancies, needs to be pursued further.

Measures of Family History of

Alcohol Problems

Shiffman (1989) has indicated that in addition to assessing factors that are relatively proximal in time to a relapse episode (e.g., temptation and confidence levels), a comprehensive assessment should also measure factors in the individual's life that are more distal, both in time and influence on drinking. These distant, often relatively enduring and unchanging personal characteristics, may provide the context that predis poses individuals toward involvement with alcohol, differing patterns of drinking, and potentially increased risk of relapse. From a clinical perspective, focusing on such distal background factors may help to predict who will relapse but not when they will relapse (Shiffman 1989).

A potentially important background characteristic is a positive family history of alcoholism, which may represent such a predisposing variable (e.g., Schuckit 1991; Tarter 1991). This may influence the nature and strength of alcohol-related expectancies and have an interactive effect on drinking behavior among young adults (e.g., Brown et al. 1987b; Mann et al. 1987; Sher et al. 1991). It may also be a contributing factor to an alcoholic subtype having a significantly different developmental course, patterns of drinking and related problems, and poorer treatment prognosis than others without a positive family history of alcoholism (Babor et al. 1992a, b; Litt et al. 1992).

The history of alcohol use in the family has been determined primarily from individuals' self-reports concerning the drinking behavior and consequences of their parents or close relatives. In some cases, this has involved the use of structured diagnostic interview protocols, such as the Family History-Research Diagnostic Criteria (FHRDC-Endicott et al. 1975), which focuses on parental drinking behavior to determine whether the diagnostic criteria of alcohol abuse or dependence are met.

Recently, a number of relatively brief and reliable self-report forms have been developed to assess familial alcohol problems. One such measure is the Family Tree Questionnaire for Assessing Family History of Alcohol Problems* (FTQ-Mann et al. 1985). The FTQ is an easily administered questionnaire that provides subjects with a consistent set of cues for identifying blood relatives with alcohol problems. They are given a family tree diagram that includes first-degree (parents, siblings) and second-degree relatives (grandparents, aunts, uncles). To assure comparability in classifying relatives with respect to their drinking, individuals are provided with descriptions of four possible drinker categories: (1) never drank (a person who never consumed alcoholic beverages), (2) social drinker (a person who drinks moderately and is not known to have or have had an alcohol problem), (3) possible problem drinker (a person who the individual believes or was told might have (had) an alcohol problem but where there is a lack of certainty), and (4) definite problem drinker (only those persons either known to have received treatment for an alcohol problem or to have experienced several alcohol-related consequences).

The reliability of FTQ classifications of paternal and maternal first- and seconddegree relatives of alcoholic and nonalcoholic subjects were examined. Both alcoholic and nonalcoholic subjects could reliably classify their relatives as alcoholics or problem drinkers over a 2-week test-retest interval

(Mann et al. 1985). Similar high levels of test-retest reliability were found in classification of family members over an approximately 4-month interval (Vogel-Sprott et al. 1985). Using liberal criteria (e.g., relative known to be a problem drinker) provided a more sensitive basis for the diagnosis of relatives' alcohol problems than more stringent criteria (e.g., relative definitely an alcoholic with reported consequences or prior treatment) (Mann et al. 1985).

Evidence for the questionnaire's validity derives from the fact that alcohol abusers had a higher number of alcohol-involved relatives than non-alcohol-abusing subjects. Alcoholics in treatment with a positive family history of alcoholism, as assessed by the FTQ, had an earlier onset of drinking, higher indices of quantity and frequency of drinking, a greater preoccupation with drinking, a more sustained drinking pattern, more serious negative psychosocial consequences from drinking, and a greater reliance on alcohol to manage their moods than those alcoholics with no familial alcoholism (Worobec et al. 1990).

A second set of measures of familial alcohol problems is based on an adaptation of the Short Michigan Alcoholism Screening Test (Selzer et al. 1976). These scales, the Adapted Short Michigan Alcoholism Screening Test for Fathers* (F-SMAST) and Mothers* (MSMAST), were developed by Sher and Descutner (1986). The individual is asked to respond to each of the 13 items of the SMAST with respect to either father's or mother's drinking behavior and alcohol-related negative consequences, with a dichotomous response format (yes/no). Separate forms are provided for the assessment of each parent, with appropriate modifications in the wording. Individuals are also asked to make a global judgment concerning whether their father or mother is (was) an alcoholic.

Overall, these scales showed a relatively high level of reliability, defined as the extent of agreement between sibling pairs on each

item. Agreement was higher for those items asking about specific behavioral acts or consequences (e.g., seeking help, being arrested); lower levels of agreement were found on items that required an inference (e.g., the presence or absence of guilt about drinking, what others thought about the parent's drinking). Reliability also appeared to be higher for ratings of fathers' drinking than of mothers.

Crews and Sher (1992) replicated this finding with a larger sample. They also replicated the previous finding that a cutoff score of 5 to define parental alcoholism was best in terms of maintaining a high level of intersibling agreement. They found a high degree of test-retest stability and internal consistency, a high level of agreement between the F-SMAST or M-SMAST and the individual's responses to the FH-RDC about each parent's drinking, and a high correlation between the individual's scores on the F-SMAST and M-SMAST for each parent and the parents' actual scores when taking the SMAST about their own drinking behavior. Parental history of alcoholism, as measured by these adapted SMAST scales, appears to be an increased risk factor in the subsequent diagnosis of alcohol disorders (Kushner and Sher 1993) and to interact with personality factors to form different subtypes of drinking disorders among young adults (Martin and Sher 1994).

Extra-Treatment Social Support

An important area to consider as part of the assessment process is the extent and nature of the individual's social support system. Perceived social support may moderate the relationship between a positive family history of alcoholism and the development of alcohol problems (Ohannessian and Hesselbrock 1993). Litman (1986) noted that the ability to access social support was one of the main methods of avoiding relapse as assessed by the Coping Behaviors Inventory.

Social skills training programs, often incorporated into the treatment for alcoholism, are thought to operate in part by enhancing the client's social support for sobriety and providing more appropriate alternatives for coping with interpersonal stress than drinking (Monti et al. 1994). The nature of social support and the level of the individual's investment in it also appear to interact with different types of treatment to effect differential outcomes, suggesting the possibility of using the domain of social support for the purpose of treatment matching (Longabaugh et al. 1995a).

Much research has examined the role of general social support in the recovery process. However, a number of authors have questioned whether this is the most appropriate focus (e.g., Beattie et al. 1993; Havassy et al. 1991). A more critical variable to assess may be the degree of support the social network provides specifically for abstinence versus continued drinking. Beattie et al. (1993) suggested that general social support is most likely to affect the individual's sense of subjective well-being, while alcohol-relevant social support is more directly related to alcohol involvement. Havassy et al. (1991) noted that both social integration and abstinencespecific functional support are important in predicting relapse.

Longabaugh and colleagues developed a family of measures designed to assess different areas of alcohol-specific social support. They separated the influence of individuals in the client's work environment from the support provided by family and friends. The measure derived to assess the former is Your Workplace* (YWP-Beattie et al. 1992). The YWP is a brief, 13-item self-report measure that can be administered either as an interview or a self-administered scale. YWP was developed from the responses of alcoholics in treatment and has three factor-analytically derived subscales: adverse effects of drinking on work performance, cues and support for consumption, and support for abstinence.

The reliability indices of these three subscales ranged from 0.61 to 0.78. The YWP scales were unrelated to measures of general workplace support, while the scales assessing adverse effects of drinking on work performance and support for consumption were related to concurrent measures of drinking behavior. The YWP scale assessing support for consumption was related to higher numbers of drinks per drinking day and the number of heavy drinking days during months 7-12 following treatment, while the support for abstinence scale was related to lower levels of drinking on drinking days. However, none of the indices of general workplace support predicted drinking behavior following treatment.

Important People and Activities (IPA) was developed to assess alcohol-specific social support from family and friends (Beattie et al. 1993; Clifford and Longabaugh 1993; Clifford et al. 1992; Longabaugh et al. 1993, 1995b). It is an interviewer-administered instrument that provides information about those individuals with whom clients have frequent contact, how important each of these individuals is to the clients, how much they like each of these individuals, and how these individuals respond to the clients' drinking and abstinence. Clients also rate the drinking behavior of those important individuals in their social network as well as how often these individuals drink during activities that are important to or valued by the client.

The scale is meant to tap three primary domains: (1) attitudinal and behavioral support from members of the social network for drinking, (2) the lack of sanctions against drinking, and (3) attitudinal and behavioral support for abstinence. The Chronbach alpha coefficient of internal consistency for the nine items included in these three indices is 0.66. (Beattie et al. 1993).

An index of affiliative support for alcohol involvement versus abstinence has also been developed (Longabaugh et al. 1993). Individ

uals characterized as having interpersonal networks supportive of alcohol involvement have important people who are perceived as more accepting of the clients' drinking and who are more likely to be drinkers themselves. Conversely, those characterized as having a network supportive of abstinence have important people who are less accepting of the clients' drinking and are more likely to be abstainers themselves.

Beattie et al. (1993) found that this index of affiliative support for alcohol involvement correlated significantly with a similar index of workplace support for alcohol involveinent as measured by the YWP. However, the IPA index of support for drinking did not correlate significantly with actual pretreatment drinking behavior.

Longabaugh et al. (1993, 1995b) found that three different forms of alcoholism treatment had differential outcomes as a function of the nature of the clients' alcohol-specific social support and the investment in this support network. At the 18-month followup (Longabaugh et al. 1995b), those subjects who had either a network that was unsupportive of abstinence or a low level of investment in their network had better outcomes following an extended relationship-enhancement therapy. A broad-spectrum treatment approach was most effective with clients who had both a social network unsupportive of abstinence and a low investment in their network or with clients who were highly invested in a social network that was supportive of abstinence. Clearly, a therapeutic focus should be directed toward the enhancement of interpersonal relationships, the development of a social network supportive of abstinence, and a means of facilitating the client's investment in this group. While this seems like a straightforward goal, it is typically underemphasized in the treatment process (Beattie et al. 1993).

A related measure is the Significant Other Behavior Questionnaire (SBQ-Love et al. 1993). The scale was developed to assess the

responses of a single significant other to the presence or absence of drinking in alcoholinvolved clients. This 24-item questionnaire uses a 5-point response scale for the client to rate the likelihood that a significant other would respond in a variety of ways to the client's drinking. Two forms are available for rating the significant other's behavior from either the client's or the significant other's point of view.

Four factors were derived for both the patient form and the significant other form of the SBQ. On the patient form, these included the perception that the significant other punishes drinking, supports sobriety, supports drinking, and withdraws from the patient when drinking. Internal consistency indices for these four subscales ranged from 0.75 to 0.87. The same pattern of factors and item loadings on factors were found on the significant other form.

With the exception of the scales measuring perceived withdrawal from the patient when drinking, the remaining scales of the SBQ showed fair concordance between the patient and corresponding significant other scale scores. General social support from family and friends was not related to the rated support of the significant other for drinking or sobriety. However, the SBQ scales also demonstrated a relative independence from measures of drinking behavior and sobriety.

Multidimensional Assessment

Measures

Drinking behavior and alcohol problems are multidimensional. As such, it is often important to have an overview of the parameters of drinking, the expectancies that accompany and potentially maintain alcohol use, and the biopsychosocial aspects of the individual's life that are affected by drinking (Donovan 1988). Assessments need to be relatively broad to capture the extent and complexity of these multiple facets. This can be