not used alcohol before the intervention or who had used alcohol only under adult supervision.

While Project SMART (Hansen et al. 1988), AMPS (Dielman et al. 1989), and the LST program (Botvin, Baker, Dusenbury et al. 1990; Botvin, Baker, Filazzola et al. 1990) have demonstrated some positive long-term results, other recent controlled research that has investigated programs focusing on resistance skills has shown little or no long-term impact on alcohol use. For example, an evaluation of students who participated in Project ALERT (Adolescent Learning Experiences in Resistance Training) found some reduction in alcohol use 3 months after the program but no differences between the intervention and control groups on use at 15 months (Ellickson and Bell 1990).

Evaluations of Project DARE (Drug Abuse Resistance Education), a widely publicized schoolbased program in which instruction is delivered (primarily to fifth and sixth graders) by uniformed police officers, have also shown mixed results (Clayton et al. 1991; Ringwalt et al. 1991). Project DARE content, which was derived from the two components of Project SMART (Hansen et al. 1988), includes both affective education (developing self-esteem and decisionmaking skills) and peer resistance training (Ringwalt et al. 1991).

An evaluation of students in North Carolina immediately after participation in Project DARE (Ringwalt et al. 1991) indicated that, although the program had affected such outcomes as selfassertiveness, attitudes about substances, perceptions about drawbacks associated with alcohol, and beliefs regarding peers' attitudes about drugs, it had no effect on students' use of or intent to use alcohol. A long-term controlled study in Kentucky (Clayton et al. 1991) also found no overall impact of Project DARE on alcohol use in a 2-year followup of students, but subgroup analyses did suggest that the program had more impact on alcohol use for students rated high on a measure of sensation seeking (the degree to which an individual seeks stimulation, novelty, and excitement and is susceptible to boredom) than for students rated low on this measure.

In the Adolescent Prevention Trial, Hansen and Graham (1991) compared resistance training with "normative education." The latter attempts to correct erroneous perceptions about the prevalence and acceptability of alcohol use and to promote conservative attitudes about use that exist among peers. Both normative education alone and a combined normative education and resist

ance training condition had significant effects on alcohol use at 1-year followup, but resistance training alone had no effect. The authors suggested that the positive findings of previous studies of resistance training programs may be due to the fact that many of these programs have included normative education components (Hansen and Graham 1991).

Hansen's (in press) review of research on prevention programs published from 1980 to 1990 indicates that comprehensive programs-programs that include a broad range of prevention strategies have also had consistently positive results. However, because only a small number of such programs have had adequate evaluations, and most of them included social influences components, specific program attributes responsible for the comprehensive programs' success cannot be identified (Hansen in press).

Few prevention projects have designed and tested programs for high-school-age youth, although these would seem warranted given that driving under the influence of alcohol is a substantial problem among older youth.

One recent comprehensive program is the Midwestern Prevention Project, or Students Taught Awareness and Resistance (called STAR in Kansas and I-STAR in Indianapolis) (Pentz et al. 1989). The STAR intervention model involves the gradual implementation of school-based resistance training, homework sessions that include interviews and role-plays with parents and other family members, mass media coverage of the program, community organization, and policy. At the time of an evaluation of Project STAR by Pentz et al. (1989), only resistance training, homework sessions, and mass media coverage had been implemented. One-year followup data indicated alcohol use prevalence rates of 12 percent for control schools compared with 9 percent for intervention schools, a 25-percent reduction.

Few prevention projects have designed and tested programs for high-school-age youth, although these would seem warranted given that driving under the influence of alcohol is a substantial problem among older youth. In one project, students who received training on how to

intervene in social situations to prevent drunk driving by peers reported more such interventions during a followup period than control subjects, who had received only an informational alcohol education program (McKnight and McPherson 1986). Changes in alcohol consumption among the trained students or their peers were not investigated in the study. An evaluation of a program focusing on personal risk identification by students and instruction for students on how to avoid situations associated with driving while intoxicated indicated that the program was not successful in preventing drunk driving (Farrow 1989).

Available data suggest that school-based programs using a social influences (e.g., resistance training, normative education) or comprehensive approach are most likely to have an impact on alcohol use of young people (Hansen in press). However, methodological problems, and in particular the fact that many evaluations of schoolbased programs have involved too few students to detect significant effects of interventions, make it difficult if not impossible to draw firm conclusions about the impact of various prevention strategies (Hansen in press). The modest effects or lack of impact on alcohol use observed in many studies of school-based prevention programs may not reflect inadequacies in underlying theory, but rather difficulties in applying theory to program and study design.

School-Based Programs That Include a Focus on Families

Interventions attempting to address familylevel influences on alcohol abuse vary in focus. These interventions have included programs that directly involve parents in school-based activities, programs conducted by parents for other parents, and programs conducted for parents by professionals.

A review of existing research on parent involvement in school-based programs (Hawkins et al. 1991) reported involvement rates of 40 to 80 percent. The highest rates were obtained when parental involvement was limited to completing homework assignments (Pentz et al. 1989). When more active parent involvement was required, participation rates were lower. To date, studies have not been conducted to determine the impact of parental involvement in school-based programs on children's alcohol use. During the 1980s, thousands of "grass-roots" groups conducted by parents were formed

across the country to address the problems of youth alcohol and drug use. A process evaluation of such groups, performed by Klitzner, Bamberger et al. (1990), suggested that pressure from parent groups was a key impetus for changes in law enforcement and school responses to alcohol and other drug abuse. A preliminary outcome evaluation by Klitzner, Gruenewald et al. (1990) also found that both parents and children in communities with parent groups reported smaller increases in children's use of alcohol over time than parents and children in matched comparison communities. Parents and children in communities with parent groups were also more likely to report lower rates of decrease in the amount of time parents spent talking with their children and the amount of time parents and children spent together, as well as increased strictness of parent control.

Programs that provide professional guidance for parents are predicated on the assumption that all parents can benefit from improved understanding of child development and improved skills in communication, discipline, and family management. In the Parent Communication Project (PCP), Shain et al. (1990) evaluated a parenting course based largely on Parent Effectiveness Training (Gordon 1970). The PCP course appeared effective in producing positive changes in parents' skills and in children's perception of family life, although parents' skills had deteriorated by the 2-year followup (Shain et al. 1990). While PCP had no overall effect on children's alcohol use, outcome data indicated that alcohol use patterns of children whose parents participated in PCP were more likely to mirror the drinking patterns of their parents than the patterns of children whose parents were in the comparison group. Specifically, when parents in the intervention group were abstainers, their children were more likely than comparison children to abstain; when parents in the intervention group were drinkers, their children were more likely than comparison children to drink.

In summary, while very few parent programs have been evaluated, some have demonstrated positive effects. Although programs for parents have theoretical appeal, available data indicate that parents are much less likely to participate in programs requiring their active involvement than in those requiring only limited involvement. Very little research has focused on the effects of parent programs on children's behavior.

Programs and Campaigns Among the General Population

Along with school-based intervention programs, a number of prevention campaigns that target broader audiences with informational messages have been developed and evaluated. In a North Queensland study (Barber et al. 1989), regular drinkers were involved in a field experiment to assess the impact of a television commercial based on controlled drinking principles. Baseline measures were taken approximately 12 months before the start of the campaign; posttest measures were taken 3 weeks after the campaign. Findings indicated that the commercial reduced intake when subjects also received a form letter alerting them to the start of the campaign.

In a New Zealand study (Casswell et al. 1990), a quasi-experimental design was used to evaluate the Community Action Project, a program for preventing alcohol-related problems. Two cities were exposed to a mass media campaign focusing on broad lifestyle goals, two cities were exposed to both the media campaign and a community organization strategy, and two cities were included as controls. The findings indicated that in communities that received the combined media and community organization campaign, attitudinal changes occurred in the direction of greater belief that alcohol can be harmful to health and lesser belief that "The secret of a good party is to keep the alcohol flowing" (Casswell et al. 1990, p. 14). Other results indicated that public support for alcohol control policies (restrictions on advertising and availability and support for price controls) declined in the control communities but remained stable in both the media-only and the media-plus-communityorganization communities.

Worden et al. (1989) developed a prevention campaign aimed at training individuals to estimate their own blood alcohol concentration (BAC) for a given amount of alcohol consumed. BAC wheels and cards were distributed via licensed alcohol outlets by retail servers trained to demonstrate their use to customers. Three communities were used to evaluate the intervention: one received the BAC wheels plus a series of televised prevention messages, one received the televised messages only, and the control site received no prevention activities. Results from nighttime roadside surveys of drivers conducted after the interventions found that the ordering of effects was as hypothesized. The proportion of drivers with BACS above .05 per

cent was highest in the control community, medium in the television-message-only community, and lowest in the community that received both the BAC wheels and the television messages.

A study by Myers et al. (1991), however, indicates that providing information that helps drinkers estimate their BACS may not necessarily lead to lower consumption. Although small in scale, this study of 30 California college students found that drinking in a student tavern was greater among experimental subjects who had been given BAC estimation charts than among controls who were not given such charts.

Research on Warning Labels

Alcoholic beverage containers were required to carry a waming label beginning in 1989 (Public Law 100-690, 1988). The text of the label reads GOVERNMENT WARNING: (1) According to the Surgeon General, women should not drink alcoholic beverages during pregnancy because of the risk of birth defects. (2) Consumption of alcoholic beverages impairs your ability to drive a car or operate machinery, and may cause health problems.

Studies that evaluate the design of the label (see figure 1) and that assess the impact of the label are beginning to appear in the published literature, but the picture these studies give at present is far from complete.

There is substantial agreement that, as of mid1990, many adults had yet to notice that the label exists. Kaskutas and Greenfield (in press) found that 27 percent of adults reported being aware of the label in a nationally representative telephone survey conducted during the summer of 1990. Gallup Poll data taken in May 1990 indicated that 35 percent of respondents said that warning labels were either “somewhat likely" or "very likely" to exist (Mazis et al. 1991). Awareness of the label appears to have grown slowly over time, as indicated in a nine-wave series of telephone surveys conducted in Utah between. April 1989 and April 1990 (Mayer et al. 1991; Scammon et al. 1992). Later data were collected for some of these survey series, but results reflecting those data have yet to be published.

Label design and placement may be factors in these relatively low rates of awareness. Laboratory studies have sought to identify design factors that would improve the noticeability of the label. A study of variations in the format and placement of the warning label on actual containers of alcoholic beverages has indicated that

warnings printed on the front of the container and warnings printed horizontally were noticed more readily than warnings printed on the back or sides or printed vertically. This study also indicated that noticeability was related to length (in cm) of the phrase "Government Warning,” number of letters in the line immediately below the line containing "Government Warning" (a measure of the overall length of the label), number of information-containing areas on the bottle, size of surface area used to print the warning, and number of lines used to print the warning (Godfrey et al. 1991). Experiments with computerproduced images of mock labels found that noticeability was enhanced by the inclusion of a pictorial or an icon and that warnings printed in a contrasting color were more readily noticed (Laughery and Young 1991a; Young 1991). A study that tracked the eye movements of subjects found that the inclusion of a pictorial, an icon, or contrasting color alone did not improve noticeability; however, the combined inclusion of all three did (Laughery and Young 1991b).

Other findings on label-related topics have also been published. A baseline survey indicated that several months before the label appeared, an overwhelming majority of respondents endorsed as "probably true" or "definitely true" each of the hazard statements in the label text (Kaskutas and Greenfield 1991); therefore, some degree of knowledge about these hazards was widespread before the label appeared. The same survey indicated that public support for the warning label policy was very high.

Less certain, in the studies published to date, are results regarding behavioral or perceptual change attributable to the label. An analysis of comparable Gallup Poll data by Mazis et al. (1991) showed a pattern of stability in perceptions of risk. Between 1989 (prelabel) and 1990 (postlabel), the proportion of adults who rated the consumption of alcoholic beverages as "very harmful" did not change significantly. There was also little evidence of change in the perceived hazards of drinking in the Utah survey series (Mayer et al. 1991). No change was found for any of four variables: the percentage of traffic accidents the respondent rated as the result of alcohol consumption; the percentage of birth defects the respondent rated as caused by alcohol; the perceived likelihood that a "regular drinker, compared to a nondrinker," would experience a traffic accident; and the perceived likelihood that a "regular drinker, compared to a nondrinker" would deliver an impaired infant.

In a preliminary comparison of 1989 (prelabel) and 1990 (postlabel) telephone survey data, Kaskutas and Greenfield (in press) found no change in the proportion of respondents who reported any of the following behaviors: driving after having had too much to drink, deciding not to drive because one had been drinking, deciding to limit drinking because one expected to drive afterward, and limiting drinking because of concern for health. The percentage of respondents who reported that they had used machinery after drinking actually increased by a small amount between 1989 and 1990.

Further analyses on many aspects of the label are pending, including analyses of additional sets of survey data that were gathered later than those discussed here. Current knowledge in this area should be considered preliminary and tentative.

Prevention Research With
Special Relevance for Women

For many years, the alcohol-related problems of women received limited attention, perhaps because they were far less prevalent than men's. Recently, a growing number of studies have found that women may be at higher risk of developing problems at lower levels of consumption than men and that the symptoms of alcohol problems, as well as their social and medical sequelae, differ between male and female drinkers, thus suggesting the need for intervention strategies tailored to women (Lex 1990) (see Chapter 8, Effects of Alcohol on Health and Body Systems).

Alcohol-related problems more prominent for women than men include serious reproductive and sexual dysfunctions; rapid development of dependence; more serious liver disease among those who are patients; victimization by others, particularly spouses; and sexual victimization.

Alcohol-related problems more prominent for women than men include serious reproductive and sexual dysfunctions (Gavaler 1991; Mello 1988; Wilsnack 1984); rapid development of dependence ("telescoping") (Anglin et al. 1987; Hser et al. 1987); more serious liver disease