Alcohol Abuse Campaigns

The same factors generally found by Mendelsohn (1973) to characterize ineffective public education campaigns permeate campaigns about alcohol as well. Haskins (1969), for example, reviewed studies on the development and evaluation of mass mediated campaigns aimed at drinking and driving (among other safety campaigns). He found the research on which campaigns and their evaluations were based to be scanty. The formative research² that was available relied solely on laboratory experiments and verbal measures. Evaluative studies were found to lack control groups and statistical analyses. He found that the message strategies relied heavily on negativistic, threatening fear appeals—an approach that may not necessarily be the most successful. ³ Haskins reaches the same conclusion as Mendelsohn: campaigns should make “systematic use . . . of known communications research, accompanied by appropriate pre-testing research at various stages of the development process” (p. 65).

Concerning the formation of alcohol abuse campaigns, only two areas have been studied (Blane 1976). One seeks to ascertain how receptive different segments of the potential audience would be to various messages pertaining to alcohol use. The National Highway Traffic Safety Administration (NHTSA) tried to assess the extent and conditions of adolescent and adult involvement with alcohol. Those involved in social or business occasions in which alcohol was served at least once in the last 3 months were considered at highest risk for being in a drinking-driving situation: they comprised 54 percent of adults, according to the study. Of these, 43 percent (about 23 percent of the total sample) felt that they may have been driving while intoxicated once in the previous year. Of these people, 75 percent had taken some action to prevent the situation from occurring.

From survey data, Grey Advertising (1975a) was able to identify and classify four basic groups according to the degree and nature of countermeasures they were willing to perform to evade or prevent drinking and driving. “Social conformers” are primarily upper-class persons who would take the least obtrusive actions (e.g., serving food at parties and/or driving intoxicated guests home). “Cautious pre-planners” tend to be older women of lower socioeconomic status (SES) and are more likely to prevent drinking and driving. “Legal enforcers” were found to be older, middle-SES people who would be likely to use legal countermeasures, such as calling the police, as a way of attempting to prevent intoxicated people from driving. “Aggressive restrainers” tend to be younger men who would be willing to physically restrain their friends from driving under the influence of alcohol.

In their first evaluation of the formative data, Grey Advertising (1975a) recommended that instead of using a shotgun approach to their campaigns as they had done in the past. NHTSA would do better to aim at drinking adults, rather than the entire population. They further recommended that NHTSA develop different strategies to sanction giving and receiving help in alcohol-related situations for “social conformers” (motivating conformity) and “aggressive restrainers” (motivating friendship).

In a second evaluation, Grey Advertising (1975b) suggested a basically similar approach for adolescents. NHTSA messages should be aimed at drinking youth as well as those adults in a position to influence them. Grey also advised that alcohol-related messages should attempt to establish a peer group norm that to give or receive aid when one is impaired by alcohol is acceptable behavior.

In another study, the Addiction Research Foundation utilized sample surveys in Ontario, Canada, in 1975 to find how much concern there was about alcohol-related problems and what remedies the public perceived to be adequate (Gillies 1975, Gillies et al. 1976a,b,c). A “Social Policy and Alcohol Abuse Survey” was conducted in order to “ascertain the attitudes of a cross-section of Ontario adults toward existing and hypothetical regulatory measures related to the use of alcohol” (Gillies 1975, p. i). In general, the survey discovered a high degree of concern about the incidence and consequences of alcohol abuse and a variety of opinions about what should be done about it.

There has also been some formative research on the development of messages for alcohol-related campaigns. Flynne and Haskins (1968), for example, wanted to find out if the statement of behavioral intention (confidence in driving after drinking) was indicative of actual willingness to perform the behavior. With such knowledge, campaign evaluations could be designed to determine whether messages aimed at reducing the unwarranted confidence of an intoxicated driver in his or her ability to drive were successful. Flynne and Haskins found that real-world observation (rather than laboratory testing) of drinking behavior was necessary for effective pretesting of messages. Real-world driving behavior differed significantly from behavior manifested under experimental conditions; hence, any means of measuring the effects of the campaign other than direct observation of real-world behavior would be invalid.

This finding applies to a program that was designed to eliminate indifference about drinking-driving messages; it resulted in the film, “A Snort History.” The didactic approach that had characterized most such messages was avoided, because tests had shown that high-fear appeals had a boomerang effect. The 6-minute film told how alcohol affects judgment in that drinkers tend to become more optimistic about their driving ability precisely when pessimism would be more prudent (Mendelsohn 1973).

Of those viewers tested, 43 percent reported the film left them feeling concerned about the effects of alcohol on driving; 30 percent said that they would consider changing their ideas regarding safe driving (Mendelsohn 1973). Real-world observation would determine how much of this cognitive shift converts directly (without further training) into changed behavior.

Admittedly, these are less positive results than those from the “National Driver's Test.” “A Snort History” does show, however, that information can be transmitted through different media with some positive effect, holding the attention of the audience while getting the message across. (In fact, the quality of “A Snort History” was good enough to have it placed as a short subject in a Denver first-run motion picture theater, where it was not subject to diffusion of impact by conflicting messages.) There may be a study or two that further refines what has been determined by these studies; aside from what has been presented here, however, there has been no further analysis of—to paraphrase Lasswell—what do you want to say to whom, how effectively do you want to say it, and through what channel.

Several assessments of program evaluation techniques (Douglas 1976, Driessen and Bryk 1972, Kinder 1975) have criticized the lack of methodological rigor. Some of the problems mentioned are: overly simplistic before-and-after designs; inability to control for confounding factors; the use of a variety of techniques in one campaign so that the effects of any one could not be singled out; and failure to consider the possible impact of a Hawthorne effect⁴ among people being studied.

Blane (1976) found that few evaluations of public education campaigns related to alcohol abuse have been methodologically sound. Typically, a simple before-and-after design is used to assess changes in attitudes, knowledge levels, and behavior of a target audience. Changes are attributed to the intervention of the campaign, but due to the typical lack of control groups in the experiments, conclusions are weakened because of unmeasured factors that may have confounded the results.

Blane and Hewitt (1977) have listed the following deficiencies in campaign evaluation strategies:

(1)

Lack of integration of evaluation into the overall campaign design, which prevents a symbiotic relationship between evaluators and designers of the campaign. A two-way flow of information is necessary so the program can be developed based on what is known from research and so feedback from the evaluative process can guide future refinement of the campaign.

(2)

Lack of sufficient precampaign testing of message content and appeal, which should be conducted with suitable samples from within the target audience.

They make the following recommendations for improvement:

(1)

Statistical analyses of results rather than merely reporting before-and-after percentages of variables as being something meaningful.

(2)

Close coordination of campaign objectives and evaluation design.

(3)

Long-term evaluation to assess other than immediate campaign effects.

(4)

Use of unobtrusive and nonreactive measures.

(5)

Determination of actual exposure of the target group to the campaign as well as correlation of these data with data concerning changes in attitudes, knowledge, and behavior.

(6)

Attention to determining the possible negative, counterproductive side effects of the campaign.

Given the lack of both formative research and sufficient evaluation, it is no wonder that previous public education campaigns aimed at reducing the incidence of alcohol abuse have had such inconclusive results. Proper use of the mass media for effective dissemination of messages is a multifaceted process that requires a great deal of planning, evaluation, and willingness to replan during the campaign if necessary.

I have shown that an integrated approach can be helpful in certain campaigns. Rather than focus on other alcohol-specific campaigns to see how they might have been changed to be more effective. I next review the elements of a well-designed campaign and what can be done to best implement them. I pay particular attention to health-related campaigns in which these factors have been manipulated to advantage.

Models of Effects

If we have learned anything from the experiences of those who have attempted to implement campaigns in the past, it is that good intentions and the utmost earnestness of convictions are insufficient catalysts for engineering meaningful change in behavior. We have had no adequate way of determining whether previous educational strategies were appropriate, as good research designs for evaluation are “a recent and still rare phenomenon” (Blane and Hewitt 1977, p. 15). Similarly, there has not been a rigorous application of the principles established by either social scientists in general or communications scholars in particular. Despite Mendelsohn's contention that it was the neglect of these principles that proved to be the undoing of much of the public information campaigns of the past, Blane (1976) reminds us that “the potential benefits of applying social science theory and methods to an understanding of the conditions under which public information and education messages (about alcohol use) are most effective in attaining social goals have been little explored” (p. 540). So, if using what is known can be effective (Mendelsohn 1973), and if previous campaigns aimed at moderating the use of alcohol have not utilized what is known (Blane 1976, Blane and Hewitt 1977, Goodstadt 1978, McGuire 1974), then there has not been an adequate implementation of communications principles (coupled with equally rigorous evaluation) to determine whether a well-conceived public education campaign to moderate the abuses of alcohol might have more positive results. I consider in this section an integrated community organization plan as a model for the dissemination of information about how to moderate alcohol-related behaviors. The usefulness of the model depends on the extent to which mass media, existing networks of social organization, and specifically created face-to-face instructional facilities are integrated. All the knowledge necessary for the immediate mass implementation of such a campaign with a guarantee of success is not currently available. What follows is a set of guidelines based on the most promising research on how to implement a community-based strategy of health education. Such a program is likely at least to provide deeper insight into how to more successfully counteract alcohol-related health problems. The Stanford Heart Disease Prevention Program 3-Community Study, described in detail below, is an example of a successful community-based health education program.

The theoretical model for the 3-Community Study was based on the work of Cartwright (1949). He argued that for mass education to be effective, three kinds of changes must occur: changes in cognitive structures (what people know), changes in affective structures (what people want to do), and changes in action structures (what they actually do). The cognitive component (which includes attention, knowledge, information, belief, awareness, comprehension, learning, etc.) refers basically to how the focus of attention—the “attitude object”—is perceived. The affective component (including conviction, interest, desire, yielding, evaluation, etc.) pertains to a person's subjective like or dislike of the attitude object. The action, or conative, component (such as intention, behavior, adoption, etc.) refers to a person's gross behavioral tendencies toward the objective in question (McGuire 1969, Ray 1973).

Cartwright believed that the mass media are usually most effective in initiating change in cognitive and affective structures; they are not often successful, however, in initiating change in action structures.

Although the existence of these three components of a “hierarchy of effects” of any persuasive message is well established, it is of critical importance to determine which particular ordering of the three components will yield the best results.

Of the six possible permutations of the three components, three orderings have dominated the literature: (1) the “learning” hierarchy of cognitive-affective-conative (K-A-B); (2) the “dissonance attribution” hierarchy of conative-affective-cognitive (B-A-K); and (3) the “low-involvement” hierarchy of cognitive-conative-affective (K-B-A). Rather than being mutually exclusive methods for affecting persuasive change, each has its place in the process of persuasion.

Initially, most people involved with public education campaigns believed that cognitive change preceded the affective, which preceded the conative (or K-A-B), in a stairstep manner; this progression was called the “learning” hierarchy of effects. Later studies, however, showed that it occurs in some situations but not in others (Ray 1973). The learning hierarchy exists primarily when the audience is actively involved in the topic of the campaign (the attitude object is already important to them) and when there are (or people are shown that there are) distinct differences between the choices offered. In this situation, members of the audience first become aware of the object; then, since it is important to them, they may develop interest, evaluate the object, try it, and, upon favorable change in attitude toward the object, adopt a new behavior.

The dissonance attribution hierarchy (B-A-K) proposes the opposite sequence of events. When a friend suggests something new, a person may try it solely on recommendation. Affect and knowledge are altered after the behavior is performed. This sequence often occurs when the source of influence is either someone to whom the person normally turns for guidance or local peer group norms. The person, involved with the attitude object, does not see that behavioral choices are available and is forced to make a specific choice of behavior as well as affective changes to support it (usually based on the behavior). Knowledge may increase on a selective basis if the person seeks out information about his or her behavior. The primary role of the mass media in this instance, after the behavior has occurred, is to reduce dissonance within the person, which may have come about as a result of the behavior and attitude changes, or to provide information for attribution of the behavior (Bem 1972, Ray 1973).

The low-involvement hierarchy (K-B-A) was explored by Krugman (1965). He noted that most people really do not care about the content of most advertising. This low involvement means that perceptual defenses to the messages are lower. Although a single announcement will probably have no effect, repeated exposure to the same message over a period of time may lead to shifts in cognitive structure. The low-involvement hierarchy is illustrated by the National Safety Council's “If you drink, don't drive. If you drive, don't drink” campaign. People had seen or heard this message so many times that other alcohol-related campaigns were thought to convey this message even though they did not discuss the issue (Harris and Associates 1974). Excessive repetition, on the other hand, may result in the message's being blocked out entirely. When there is little objective difference between alternatives, when the audience perceives little difference between alternatives, or when the audience does not care about the magnitude of difference between alternatives, this low-involvement hierarchy is most likely to be effective.

What would be the most effective approach for an alcohol-related campaign? It would be a mistake to place all the emphasis on one particular strategy. A campaign designed to change the drinking behaviors of chronic abusers of alcohol would not use the low-involvement approach, for example, because these people have rather high involvements with the object of the campaign. A campaign designed to educate children about some of the problems of overindulgence might have more success using it.

Goodstadt (1978), reviewing various studies of the learning (K-A-B) approach to influencing drug-associated behavior, found little evidence of behavioral change resulting from attitude change. The difficulties, as he sees them, are the inability to define distinctly the behaviors to which the cognitive and affective components of attitudes refer, the inability to factor out situational variables that also have a strong influence on behavior, and difficulty in measuring the attitudes themselves. He suggests that “a more effective medium for change might be via both direct and indirect behavior influence rather than through attitude change attempts” (p. 266)

Legally required or mandated behavior change, however, especially related to drug or alcohol use among adolescents, can lead to a hardening of attitudes and clandestine behavior as an act of defiance (McGuire 1974). Without appropriate persuasive communication, mandated behavior change can do more harm than good. For example, the campaigns to require people to use seat belts (Robertson 1976) have engendered hostile responses. In response to government-mandated installation of interlock systems on belts in cars, many people disconnected the belts or the warning buzzers rather than comply with the intent of the law.

Goodstadt (1978) suggests a fourth model (V-P-B), based on the belief that values influence behavior, which he believes may help decrease the incidence of drug abuse. The emphasis on values in health education is a result of dissatisfaction with traditional methods. Familiar problems, however, have been found in most drug-related value-behavior programs: “Results from the very few programs which have been evaluated are inconclusive because of (a) their scarcity; (b) their methodological problems; and (c) conflicting or ambiguous results” (p. 271).

An adequate determination of the levels of people's knowledge, attitudes, and behaviors concerning alochol is essential to determine the most effective approach.

The Stanford Heart Disease Prevention Program: An Integrated Model of Behavior Change

Attitude Composition

Rokeach (1966–1967) defined an attitude as “a relatively enduring organization of beliefs about an object or situation predisposing one to respond in some preferential manner” (p. 530). Fishbein and Ajzen (1975) add that it is the evaluative or affective nature of attitudes that distinguishes them from other concepts. Although there is some discussion involving the exact nature and measurement of attitudes (Fishbein and Ajzen 1975), there is almost universal agreement that attitudes have both a cognitive component and an evaluative component, that they are fairly well organized and long-lasting, and that there may be possible interactions between particular attitude objects and the situations that involve them. Change in an attitude indicates a change in one's predisposition to respond to a particular object in a certain situation.

The susceptibility to change of an attitude depends on the conception of how an attitude is formed. Bem (1970) and Jones and Gerard (1967) have posited that the attitude structure can be seen as a syllogism of “psycho-logic.” That is, an attitude is the conclusion of a syllogism combining a belief about an object with a relevant value. The syllogism may not be perceived by the person but rather may be an implicit part of the thought process.

For example, in the syllogism, “Cigarettes cause cancer. Cancer is bad. Therefore cigarettes are bad.” the conclusion (“Cigarettes are bad”) is derived from a belief connecting cigarettes to cancer combined with a negative value toward cancer. The syllogism may appear to be overly simplistic, since everyone agrees that cancer is bad; in fact, the “Cancer is bad” step may be left out entirely, since it is implicit in our cognitive thought process. Yet, somehow we learned something about cancer and were persuaded by it, coming to believe that it is bad (Roberts 1975). A statement like “Frequent consumption of large amounts of alcohol is related to higher levels of high-density lipoprotein” probably has no evaluative meaning for most people. Only after people are taught or persuaded about an evaluative relationship, e.g., between high-density lipoprotein and heart disease, that an attitude about frequent consumption of sizable amounts of alcohol enters their logical structure.

Many attitudes are based on a similar combination of evaluative and nonevaluative beliefs, although the linkage is not quite so obvious. Attitudes may also be conceived of as having both a horizontal and a vertical structure. Horizontally, a person may have more than one syllogistic attitude about a particular attitude object. To extend the example above, I may also believe that “Smoking pollutes the air. Air pollution is bad. Therefore smoking is bad.” The vertical structure of attitudes means that a given attitude may be a second, third, or nth derivative of other beliefs and attitudes. So, regarding alcohol consumption one might also think, “Having a six-pack every night after work means that I drink a lot of alcohol. Drinking is bad. Therefore having a six-pack every night is bad.” To complicate matters further, higher-order beliefs are bolstered by the breadth of the horizontal beliefs around them, with more than one syllogistic chain culminating in the expressed attitude (see Figure 1).

Figure 1

Hypothetical structure leading to a positive evaluation of drinking.

However, as Bem (1970) suggests, the more layers there are to the vertical composition of the syllogistic structure, the more points are open to attack. Each of the underlying premises on which the individual had built her or his attitude may afford an opportunity to precipitate some change. In the chain of attitude composition, says Bem, “higher order belief would appear to be only as strong as its weakest link” (p. 11).

As we can see in Figure 2, there are three syllogistic paths that might lead to the conclusion that drinking is good; even if one of the chains were weak and vulnerable to attack, it is supported by the others. If Figure 2 were a true depiction of a person's attitude structure about drinking, it would be possible to convince him or her that Baptist preachers are good people who may not drink for other reasons; such persuasion, however, would not change the overall attitude, since this person still wants the respect of his or her peer group and a decrease in tension at work. We can also see that this illustration depicts a “psycho-logic.” That is, the syllogisms may be internally consistent to an individual but have no basis in logic to an objective observer (assuming that objective and accurate observation of all the factors of another's attitudes were possible). Conversely, because something may be objectively true does not make it subjectively true. (Indeed, there are still some people who believe the earth to be flat.)

Figure 2

Results of project C.L.A.S.P. in Santa Clara County, California. Source: Perry (1979). Reprinted by permission.

The structure of these “psycho-logical” syllogisms helps us to conceptualize several ways to initiate a change of attitude. One might aim the persuasive appeal at a cognitive component of the attitude structure. In terms of the illustration, one might attempt to convince the person that there are many Baptist preachers who do in fact drink. One could try to modify the value premise, i.e., convince the person that Baptist preachers are good. One could try to modify the intensity of that value, i.e., show the person that although Baptist preachers may not be perfect, they are certainly not all that bad. Or one might try to circumvent the belief structure entirely and try instead to construct a new one. One might suggest new cognitions about some negative aspects of excessive drinking behavior, in the hope that the antidrinking values attached to those new beliefs would outweigh the prodrinking values of the old one.

Ross (1976) postulates that people relate to the world as amateur or intuitive psychologists. They observe events and attempt to ascribe causes to them. These causal attributions tend to influence subsequent attitudes and behaviors. This is the heart of attribution theory. This kind of causal attribution, however, often leads the attributor to a number of errors.

The most common mistakes result from the limited experience of the perceiver. A person normally has only his or her existing knowledge, attitudes, and previous attributions (which may have caused previous errors in perception) on which to base judgments. An example is a young person who sees a friend drink to intoxication, attributes this as normal (“with it”) behavior, and subsequently assumes the behavior for herself or himself. Another example is seeing the hero in a Western walk into the local saloon and down shot after shot of whiskey, then beat the bad guy and win the heroine; the audience might attribute social efficacy to the behavior and try to assume it.

One focus of a public education campaign aimed at reducing alcohol abuse could be the dissemination of messages showing that the excessive drinker is not, for example, a “he-man” or that teenage peers do not consider excessive drinking to be normal or “cool” behavior.

Social Learning Approach

In some instances, the knowledge that a problem exists is insufficient impetus to change certain health behaviors. As Cartwright (1949) indicated, it is also necessary to teach people the action structure necessary to attain the desired goal. Sometimes the psychological costs are thought to be too great to overcome. For example, many people realize that smoking cigarettes is harmful to their health, yet they continue. They may feel that the costs associated with quitting (such as “I'd put on weight,” “I would crave cigarettes all the time,” or “I just don't have the willpower”) are greater than the benefits to be derived. This requires that they have to be taught how to quit and, having quit, how to maintain their new behaviors by responding to old behavioral cues (which had been associated with smoking) in a new manner.

An interpersonal strategy for teaching behavior change has been shown to be effective in several areas: smoking (Bernstein and McAlister 1976, Thoreson and Mahoney 1974, McAlister 1978, McAlister et al. 1979); behavior associated with cardiovascular risk (Meyer et al. 1976); and to some extent weight reduction (Harris 1969, Penick et al. 1971, Stunkard 1972). The social learning approach has its theoretical foundation in the work of Albert Bandura (1969, 1977). The theory of social learning asserts that most behavior is learned from the modeling of others and the selective reinforcement of certain behaviors directly, vicariously, or symbolically. In the modeling of new behavior, people witness stimuli, responses to those stimuli, and rewards associated with the behavior. Selective reinforcement can change the internal reward value of an undesirable set of responses from high to low while establishing new patterns of behavior connected with the high value of reward.

For example, the modeling of peers is an especially strong influence on adolescents. Peer pressure appears to be the major factor in the initiation of adolescent smoking (Maher 1977, Newman 1970a,b) as well as drug and alcohol use (Braucht et al. 1973, Gorsuch and Butler 1973, Jessor and Jessor 1971). From Bandura's (1977) analysis of social learning we can see that new behaviors tend to originate from exposures to powerful models—i.e., attractive people who appear (implicitly, at least) to be rewarded by their behavior.

Those who begin smoking early are likely to have the behavior modeled for them by popular peer group members. Furthermore, the behavior is likely to be reinforced by admiration for undertaking such a “daring” experience (Schneider and Vanmastright 1974). In this situation the interaction of influences is particularly striking. Adolescents see peers modeling a certain behavior (smoking, drinking, etc.); attribute maturity or daring to the act (“he must be ‘with it' if he's doing such a thing”); and, quite frequently, fear social disgrace for not attempting to imitate the behavior.

In this case an “inoculation” strategy may be most effective in moderating behavior. This approach, outlined by McGuire (1964, 1973), is analogous to the practice of inoculation in preventive medicine. If social pressures to adopt a negative health behavior are seen as a virulent “germ,” then inoculation against “infection” (actual adoption of the negative behavior) can expose people to a mild form of the argument and teach them skills for resisting pressures to adopt the unhealthful behavior—“antibodies” (Roberts and Maccoby 1973).

This concept of inoculation or the development of counterarguments can be applied to a smoking or drinking intervention. Adolescents who are likely to be pressured by peers to drink can be forewarned of these pressures and trained to give assertive counterarguments.

For example, they can be trained to reply, “If I drink to prove to you that I'm not a chicken, what I'm really showing you is that I am afraid of not doing what you want me to do. That's really acting like a chicken. I don't want to drink.” Or, if young people are likely to see older peers showing off and acting “with it” by drinking, they can be taught to think to themselves, “If she were really ‘with it' she wouldn't have to prove it by drinking.”

To teach these counterarguments, a peer approach appears to work best. During early adolescence, peers take over from adults as the primary source of influence (Utech and Hoving 1969). Teaching by peers can efficiently provide traditional health education to large numbers of elementary school children (McRae and Nelson 1971). In addition, training young people to advise their peers about how to handle personal problems is also a successful approach (Alwine 1974, Hamburg and Varenhorst 1972).

Using “attractive” peers to teach counterarguments has been an effective strategy for the prevention of adolescent smoking behavior (McAlister et al. 1979). In Project C.L.A.S.P. (Counseling Leadership About Smoking Pressures), initiated in the Santa Clara County, California, school system during the school years 1977–1978 and 1978–1979, two teams of high school students taught classes in smoking prevention for a total of 9 days to all 7th graders in the treatment group. The high school students challenged the 7th graders with pressures and arguments to start smoking. They then modeled methods for handling these pressures through the use of movies, small-group discussion, slide shows, modeling, contests, and role-playing.

The results are noteworthy. The program began with a baseline analysis in September 1977. After 9 months, the treatment group had significantly fewer people who smoked “within the past month” or “within the past week” than the two control groups. Replication with the 1978– 1979 7th-grade class also yielded significant results in December 1978 and June 1979 (p <0.0001).

A similar program conducted in five San Francisco Bay area high schools, using the same techniques (except the older peers came from Stanford University), also showed a significant reduction of smoking behavior for the treatment groups from September 1978 to January 1979 (Perry 1979).

A pilot study of this approach to intervention in the adoption of alcohol and drug behaviors by 7th graders was also attempted. Peer leaders described situations in which a young person was being pressured to drink alcohol and smoke marijuana. After a contest was held to see who among the students could come up with the best ways to resisting pressures, they all were given buttons that read “I'm naturally high.” Peer leaders told the students that that is how college students respond to pressures to drink or to use drugs. They discussed the meaning of this phrase with the students.

Preliminary data suggest that this approach may have had some positive impact on alcohol and marijuana use: 37 percent of the students in the treatment group reported having drunk alcohol by the end of the year, while 51 percent reported doing so in the control group. Marijuana use was reported by 14 percent in the treatment group versus 25 percent in the control group (McAlister et al. 1979). However, a more extensive application along with more rigorous evaluation procedures will be necessary in order to gain a more thorough understanding of such efforts.

The social learning approach is not concerned with information per se, but rather with the particular behaviors to be changed. Initially, goals are established for the person and systems are initiated for modeling new behaviors, practicing them, and rewarding their successful practice. Ultimately, the aim is for the person to internalize the value of the behavior so that its exercise becomes self-fulfilling; at that point, the person will be better equipped to manifest the new behavior in the context of her or his environment.

Recently, Bandura has stressed cognitive aspects of learning and performance, particularly self-efficacy. People are taught incrementally that they can exert control over behavior they had previously felt powerless to affect. Each performance of the new behavior raises the level of efficacy of the person as well as the expectancy of greater efficacy in the future (Bandura 1977, 1978, 1979; Bandura and Simon 1977).

This technique has had some success in programs of weight reduction (Bandura and Simon 1977, Stunkard 1975, Stunkard and Mahoney 1974). For example, Bandura and Simon (1977) report continued weight loss after 14 weeks (p <0.001). Stunkard (1975) reports significant differences in weight loss (p <0.001) after 12 months. However Stunkard and Penick (1979) urge caution in being too enthusiastic about long-term findings as yet.

Group Influence

Cahalan (1975, p. 28) is convinced that “unless media campaigns are planned to be closely linked to people-to-people grass-roots programs to get widespread personal commitments to set an example of moderation, the campaigns will be of little consequence—especially since any campaigns for moderation in drinking have to compete against the many millions of dollars poured into advertising by the alcohol industry.” He argues that only within a “social movement toward moderation” can anything be done to reduce alcohol problems. Part of that social movement would be “directly related to setting a good parental and peer-group example [his emphasis] at home and in other social institutions” (p. 26). In his view, we should seek to intervene to initiate (or foster) such an interpersonal social movement.

The existence of such a social support system implies the necessity of proper modes of communication. Strategically placed people diffusing the right information and modeling the correct behaviors are essential ingredients in any program of social change (Rogers and Shoemaker 1971). Part of that task can be handled by the media, which can be the original disseminators of information. Mass media messages, however, are not received equally by each person. Rather there exists a multistep flow, wherein some people get their information directly from the media, others through an opinion leader who has gotten the information from the media and/or other opinion leaders, etc. (Meyer et al. 1977).

Organizations to which people belong can function as both leaders of opinion and micro-social support systems. Similarly, local social services, to which the media can direct people, can be used to introduce new information, act as a classroom to teach appropriate techniques to modify behavior, and serve as a social support system for individuals undergoing treatment. Thus, the person who comes to one of these social service organizations would begin to have a new peer group helping to reinforce her or his commitment to a new system of drinking behaviors.

Not only would these organizations have readily available information to give to interested persons who come for help, they would also need personnel qualified in a variety of group training methods aimed at the modification of alcohol-related attitudes and behaviors. Drawing from what we know of the processes of small group dynamics, we can see that an effective program for modifying drinking behaviors is possible.

There is a respected tradition of literature on group processes. Festinger et al. (1958) documented pressure toward intragroup uniformity. If a group becomes highly cohesive, Back (1958) found, members will make an extra effort to change and reach agreement. Asch (1958) also found that people feel pressure to conform to the norm of the group, thereby sticking with the program.

Lewin (1958) found that intragroup decision making and norm-reinforcing processes facilitated change more than a simple instructor-to-group talk. The initial step could be for individuals to make a pledge within the group to change their behavior to the degree mandated by the program. “Public announcement of the decision by the individual may add to subsequent resistance, and more particularly so, by taking an irrevocable action on the basis of the decision, especially if this action involves some sacrifice on the individual's part and is mandatory” (McGuire 1974, p. 12). If everyone in the group made the statement of intent to change behavior, each individual would be provided with a “health-relevant reference group in which patients [would be] able to interact with others who are in the same boat as themselves” (Stokols 1975, p. 141). The use of this technique by SHDPP met with success (Meyer and Henderson 1974, Maccoby and Farquhar 1975).

As mentioned earlier, training people to change their behavior related to alcohol is similar to training them in dietary change, in that it seeks, not necessarily to eliminate the behavior entirely, but to teach people to control and modify their behavior in such a way that the degree of risk is reduced. Examples of successful interventions using small-group interaction processes can be found in the summary report of Lewin (1958). He reports an experiment by Bavelas, Festinger, and Zander in which the group decision method was superior to lectures in persuading women to include intestinal meats in their families' diets. The same was found by Radke and Klisurich to be true for the adoption of the use of powdered milk. Lewin, Radke, and Klisurich replicated this study by persuading mothers to adopt the use of cod-liver oil and orange juice in their infants' diets (cited in Lewin 1958).

Despite objections raised about the validity of these findings (Pelz 1958), more recent research in the medical literature supports the contention that for the behaviors we seek to modify a group treatment approach will be more effective than individual treatment. London and Schreiber (1965), for example, showed that group management was better than individual management of weight control under three experimental conditions: drug, placebo, and no medication. Somewhat similar findings were found in a program to reduce cardiovascular disease risk (Meyer and Henderson 1974).

Although negative results were obtained in a smoking program (Hunt et al. 1971), more recently Heath et al. (1979) reported a higher level of success than many previous smoking campaigns attempted in Australia. Using a group therapy approach to teach nonsmoking coping behaviors, Heath and his colleagues show a minimum 65-percent success rate, defined as complete cessation being maintained for 12 months, for each of 3 consecutive years. In their last reported intervention (1978), with 370 persons treated, Heath et al. report a success rate of 71.6 percent. They attribute their success to the use of community education for a period of 4 years, rather than previously attempted experimental exercises.

Most programs of smoking cessation are successful only with a minority of smokers. Participants most often resume smoking after a few months. Heath et al. (1979) were able to achieve a higher rate of cessation of smoking; in addition, at the end of 1 year, 72 percent were still nonsmokers. Maintenance of change is the essential element in these programs.

Bandura and Simon (1977) reported success with groups of obese people in moderating eating behaviors. Here, individuals were taught to set intermediate goals for themselves as the program progressed. As goals were attained, self-efficacy grew, which increased motivation to reach higher goals, which led to further weight loss. Since this progress took place within small groups, each person's success served to suggest to the others in the program that “If she can do it, I can do it, too.”

In the Stanford Heart Disease Prevention Program 3-Community Study, programs of professionally led behavior modification to reduce the rate of smoking among high-risk individuals were implemented with a certain degree of success (a net decrease of 35 percent in number of cigarettes smoked daily after 2 years [Farquhar et al. 1977]). In the current SHDPP 5-City Project it is hoped that people can similarly be trained to teach appropriate behavior modification techniques, in order to carry on the intervention and the program of community involvement.

To apply this process to the problem of alcohol abuse, behaviors determining nonuse or nonabuse of alcohol would be specifically identified (such as how to cope at parties, business lunches, sporting events, etc.). Goals for improvement would be established and a timetable set. The instructor would model or simulate the new behaviors. For example, a party would be arranged (perhaps in a nonclinical locale for the sake of realism) and people would be shown how to cope with the social and/or internal pressures to drink. They would be encouraged to practice the new behavior and rewarded for improvement by praise, attention, a token point system, etc. The ideal behavioral outcome would be for participants to internalize the new system of behaviors so that they become self-rewarding and/or reinforced by such factors as peer approval, greater family harmony, etc. With each step each person would have a better sense of efficacy in controlling her or his own behaviors, thereby increasing the expectation of controlling them in the future. Because a great deal of alcohol abuse occurs in social situations, group counseling may be more effective than individual counseling since it gives people practice in the desired behavior in the setting in which the behavior would naturally occur.

It is necessary not only to teach change in the clinical environment but also to make the change carry over to the everyday world. The use of group learning and reinforcement is the most efficacious strategy (short of retraining everyone associated with each participant, then retraining everyone associated with them, etc.) not on an individual basis.

Source

The first independent variable is the source—the “who” in Lasswell's paradigm. Identical messages transmitted in similar fashion to the same persons may have varying degrees of persuasive impact, depending on who sends the message. Source factors are summarized by McGuire (1969) as credibility, attractiveness, and power. Despite what may appear to be obvious, these factors tend to be glossed over quite frequently.

In attempting to convince people of the correctness of a position, a source's persuasiveness depends on credibility, which in turn rests on the perception by the receiver that the source is competent, knowledgeable, and trustworthy. A number of studies have shown that differences in perceived expertise are correlated with how much opinion change a source achieves (cited in Roberts 1975). Thus, we must determine who the target population of a campaign believes to be competent and trustworthy in the area of alcohol abuse education.

There have been difficulties with studies that have tried to locate exactly what cues account for a source's credibility. Choo (1964) attributed messages about smoking and cancer to either a public health physician (high credibility) or a public relations man from the tobacco industry (low credibility). Bochner and Insko (1966) attributed messages advocating less sleep to a noted physiologist (high credibility) or a YMCA director (low credibility). Although one would like to attribute the effects to the source with higher credibility, other factors such as age, intelligence, knowledge, social status, etc., tend to confound the results. We might ask whether the testimony of an older, cured alcoholic who has had years of experience abstaining from alcohol is more effective than a younger one who just kicked the habit and has a fresher knowledge of the pain involved.

Sources who are perceived as having something to gain from the recipient's behavior are less likely to be yielded to. We may well surmise that a campaign initiated by a commercial treatment facility to moderate drinking would be less effective than one initiated by the liquor industry, because the treatment facility's campaign could be seen as motivated by profit, whereas the liquor industry's campaign, if successful, could be seen as reducing its income.

One generalization that can be made from the research on source credibility is that no single source is highly credible for all messages to all audiences. Pretesting specific messages for various segments of the target population will help to determine which sources are most likely to be well received by the audience.

Here we run into an immediate problem. If the government plans to sponsor an alcohol abuse education program, should government officials be the speakers in the messages? Indeed, should the government be identified with the campaign at all? Following what could be termed the “swine flu fiasco” of a few years ago, it would seem that the public is extremely wary of certain government information programs that might attempt to influence their health behaviors (Medical Progress and the Public, n.d.). We should not assume that government affiliation with an alcohol education program will be highly credible to the public. Thus, the issue of who is a credible source and whether government affiliation has an effect on the credibility of the message must be one of the major areas of formative evaluation.

A second source factor, attractiveness, can be seen as a confluence of the receiver's feeling of similarity to, familiarity with, and likability of the source. A good deal of research indicates that the persuasive impact of a message increases linearly as a function of the perceived similarity between recipient and source. Consistency theorists (e.g., Newcomb 1953, Osgood and Tannenbaum 1955) posit that being liked by the receiver increases the source's influence.

How the misreading of source attractiveness can cause a campaign to backfire is best illustrated by the Australian “Stop a Slob from Driving” campaign. The idea was to portray drinking drivers as slobs; since people would not want to be perceived as slobs, they would reduce their driving while under the influence of alcohol. Instead, many people found “Mr. Slob” to be funny, attractive—a person they wouldn't mind being like. The result was that this particular campaign may well have done more harm than good by strengthening existing dispositions toward drinking. A pretest of this image might have revealed this flaw in the campaign.

Using celebrities may also not be the best approach. Dana Andrews, Dick Van Dyke, Art Carney, and Don Newcombe, among others, have appeared in alcohol campaigns. Since they are reformed alcoholics they have high credibility. However, their attractiveness to the audience (especially those under 30), as well as the small degree of perceived similarity by the audience, most likely minimized the impact of their appeal.

To conclude, the sponsoring agency of an alcohol education strategy must pretest the proposed target audience to discern both the agency's credibility to the audience and the credibility, attractiveness, and power of those sources within the messages themselves.

Message

There are many aspects of message composition, including what information is included or left out, whether the message contains opposing arguments, whether the message aims at values or beliefs, whether the conclusion is explicit or left to the receivers, and how extreme a behavior change is advocated. Some general conclusions from the vast research literature in these areas (see McGuire 1969) are of particular interest for an alcohol program.

Although it was once believed that the conclusion of a message should be left to the receiver so as to involve him or her to a greater degree, there is ample research showing that including all arguments and explicitly drawing the conclusion leads to greater persuasive impact. Regardless of level of intelligence, most people seem insufficiently informed or motivated about the subjects of most campaigns to draw the intended conclusions. For communication to be persuasive, notes McGuire (1969), “it is not sufficient to lead the horse to water; one must also push his head under water to get him to drink” (p. 209).

Beyond advocating change, the message should tell the receiver how to implement the position advocated. As Cartwright (1949) tells us, “the more specifically defined the path of action to a goal (in an accepted motivational structure), the more likely it is that the structure will gain control of behavior.” Furthermore, “the more specifically a path of action is located in time, the more likely it is that the structure will gain control of the behavior” (p. 443). One of the drawbacks to most campaigns is that they have not been sufficiently specific about the course of action they advocate.

Messages should be designed to advocate specific behavior regarding a specific act. For example, “If your friend is drunk at a party, take him home in his car and ask another friend to drive behind to pick you up,” rather than “Friends don't let friends drive drunk.” In the first instance, a specific conclusion and action is advocated; in the second, the receiver is left to devise a proper course of action. Pretesting has shown that this specific solution is more palatable to people than the confiscation of an intoxicated friend's keys. Pretesting of messages is essential.

There are also sound reasons for including more than one side of a debate. People come into contact with numerous messages advocating drinking (from friends, media, coworkers, etc.). Roberts and Maccoby (1973) show how teaching people counterarguments may be effective to help them reinforce their moderate (or non) drinking behavior. The counterargument should not be so strong as to tempt people to try the behavior the message warns them against. Also, the political realities of contemporary America suggest that we eschew an aggressive counteradvertising campaign against the liquor industry. Wallack (1978) documents how, in an alcohol campaign in California, messages had to be changed after serious objections were raised by both the liquor industry and the governor.

Many alcohol-related campaigns have relied on some degree of fear appeals to convey their messages. For example, a 1971 NHTSA print ad had a headline, “Today Your Friendly Neighbor May Kill You.” The National Safety Council's “Scream Bloody Murder” campaign had such messages as “Drunk drivers add color to our highway” and “Drunk drivers bring families together. . . in hospital rooms and at funerals,” etc. The extremity of these messages may have done more harm than good. Many people who have driven after having had too much to drink know that drinking and driving does not inevitably lead to a fatal accident. Research shows that extreme fear arousal leads to avoidance responses by receivers because actions recommended in the message may not be perceived to be strong enough to cope with the degree of fear aroused. According to Chu (1966), the advantage of high-fear messages is realized to a greater degree when the recommended action to cope with the problem is perceived by the receivers as highly efficacious and the danger is immediate. In other instances a moderate level of fear arousal coupled with directions for realistic coping behaviors may be the most efficacious approach. Janis and Feshbach (1953) found this to be the case in a program to teach the causes and prevention of tooth decay, as did Leventhal and Watts (1966) in a program to induce people to get chest X-rays, and the San Antonio, Texas Alcohol Safety Action Program “Fear of Arrest” campaign (Hawkins and Cooper 1976).

There are implications of this approach for an alcohol abuse education campaign. Sternhal and Craig (1974) analyzed the fear appeals used in several health and safety campaigns. They concluded that “including specific recommendations which the audience perceives to be effective in reducing physical threat” (p. 31) is an approach that may engender a substantial degree of compliance. While they feel that the threat of physical harm will increase persuasion only if the source has high credibility, other research contends that the threat of social disapproval is more effective than the threat of physical harm.

Thus, to be effective, a message that arouses fear must be accompanied with the suggestion of a specific action or behavior for the receiver to perform. It should be something that can be done easily and immediately; the person should be able to feel a sense of efficacy in performing the suggested act and feel fairly certain that it will be successful in preventing the threatened damage. In the Stanford Heart Disease Prevention Program study, a specific suggestion, such as “cut down to 2 to 3 eggs a week,” produced greater responses in the heart disease program than vague, long-term suggestions (Farquhar et al. 1977). It would seem that an equally effective suggestion might be “Why not make every third drink a soft drink at the next party?”

The use of specific behavioral recommendations in the areas of smoking, diet, and exercise has been shown to be effective (Farquhar et al. 1977). The simple presentation of useful skills as alternatives to harmful behaviors will most likely lead to improved healthful acts. However, this is a very complex phenomenon; it is essential to pretest various appeals and approaches to the prevention of alcohol abuse to discover their effects; constructing messages should not be based only on communicator-perceived “common sense.”

A word of caution is necessary. Several high-fear messages have deliberately disseminated untrue information in the hope that the fear aroused would be sufficient to keep people from performing some behavior. Films such as “Reefer Madness” come most easily to mind. When people, especially young people, experience firsthand that smoking marijuana will not inevitably lead to heroin addiction, prostitution, dope-crazed homicide, and an eternity of sin and degradation, they will, more than likely, not heed any messages about the negative implications of marijuana in the future. Nor will they be inclined to heed the advice of the communicator on any number of topics in the future.

Similarly, many people know that drinking and driving does not inevitably lead to a fatal accident. Thus, the high fear-arousing messages connecting this behavior with gory highway or hospital scenes may have the same ultimate effect as films like “Reefer Madness.” We should be cognizant of this phenomenon in designing alcohol education programs.

Channel Factors

The part of the communication process that has received the least research attention has been the channel. What few studies exist have focused on whether a print message is more effective than a radio or television message, whether face-to-face contact is more effective than films, etc. Reports of campaigns typically devote little, if any, attention to how the choice of particular media was arrived at, how the medium was used, and what problems were encountered in dealing with people who work in the different media. From scientific perspective this lack of attention is unfortunate, because it makes replication difficult, if not impossible.

Channels are an important link in the chain of persuasion since, in the words of a radio veteran, “You can't sell them nothin' if they don't see the spot.” Typically, the credit for the success of advertising has been attributed to the artful creation of messages, jingles, etc. Little, if any, notice has been given to careful planning of the best ways to reach the target audience, both for the widest dissemination of the message and for the most effective use of the advertiser's dollar. For example, in the Stanford Heart Disease Prevention Program's 3-Community Study, radio was employed heavily for Spanish-speaking people because it was ascertained (through formative evaluation) that Hispanics in the target communities spent a great deal of time listening to Spanish-language radio.

A great deal of the effectiveness of the Stanford media campaign was in its ability to get people to think about issues that affected their health, issues about which they had probably never thought before. What change did occur after the campaign was implemented was due not only to the campaign but also to many converging factors. The campaign was most effective in setting the agenda of what was salient for people in the test area. Once people's agendas were altered, behavior change became more likely.

This agenda-setting function of the mass media has been the focus of a great deal of research attention in the past 15 years. It was best articulated by Cohen (1963), when he noted that “the press is significantly more than a purveyor of information and opinion. It may not be successful much of the time in telling people what to think, but it is stunningly successful in telling its readers what to think about” (p. 13). This function has been defined as “. . . specifying a strong positive relationship between the emphases of mass communication and the audience” (Comstock et al. 1978, p. 320).

There is ample evidence that news items that are heavily emphasized by radio, television, and newspapers become the most salient issues for the public (Becker et al. 1975; McCombs 1972, 1974; McCombs and Shaw 1972; Robinson 1972; Robinson and Zukin 1976, among others). While most research on agenda setting has focused on political information, its connection with public health information is clear. The U.S. Public Health Service reports that there has been a steady increase over the past several years of people who believe that cigarette smoking is harmful to health. Over 90 percent of those living in the three communities of the Stanford study, both smokers and nonsmokers, believed that quitting smoking would help a person live longer.

In the Stanford 3-Community Study, the information disseminated by the mass media was sufficient to induce people to change their behavior. People reduced their consumption of some high-cholesterol foods on the sole basis of what they saw on television, heard on the radio, and/or read in the local newspaper. For example, there was greater than a 40-percent reduction in the consumption of eggs over a 3-year period in the community with massive intervention and over 35 percent in the media-only community. In these communities, once the new agenda was set, the information was sufficient to effect some change in behavior in the media-only community (Farquhar et al. 1977).

Once an agenda has been set, it is possible to get people to pay attention to communications of greater depth. Thus, in the Stanford Heart Disease Prevention Program, television and radio spots got people to heed longer messages and to learn from those longer messages, e.g., advice about food purchases, cooking, etc.

Although most public education campaigns cannot mount huge, multimillion dollar campaigns due to budgetary restrictions, wise use of the various available media may make the difference between no positive outcomes and the initiation of some degree of change. The remainder of this section discusses some of the important channel factors necessary to plan a campaign.

First, different media have different capabilities. Television and radio can be used for spot announcements, public affairs programs, and entertainment programs. Mendelsohn (1973) shows that a well-planned educational program, such as “The National Driver's Test,” can be entertaining as well. The Stanford Heart Disease Prevention Program aired an equally interesting program, “Heart Health Test.” Spot announcements can call attention both to special programs and to various themes. Newspapers can be used to carry doctors' columns, to give people advice and information, and to increase their awareness about the various problems associated with alcohol consumption. In the Stanford program, billboards were used to reemphasize slogans associated with the campaign (Maccoby and Farquhar 1975, Maccoby et al. 1977) and used to communicate short messages (long copy can be dangerous for drivers to try to read); they can act as a tool of reinforcement.

Other media include newspaper stories, business cards, direct mail (Maccoby and Alexander 1980), pamphlets, newsletters, films, subway and bus cards, and the insertion of campaign materials in paychecks and family allowance checks (Blane and Hewitt 1977).

Typically, spots on the electronic media must be short (usually 1 minute on radio, 30 seconds on television) but can contain quite a bit of information. Since people can read at their own pace, newspaper and magazine ads can contain somewhat more information. As noted, billboards, bus and subway cards, and business cards (called “out-of-home” media) must be very short, with little information.

A problem usually encountered involves public service announcements. Many radio and television stations take public service announcements (PSAs) and run them at 3 a.m., when commercial time is the cheapest and often unsold. If there is no money to buy spots at more convenient times, however, all is not lost. All stations are licensed by the Federal Communications Commission (FCC) and are mandated to operate in the “public interest, convenience and necessity.” Leaving aside a discussion of whether they fulfill that mandate, they reapply for a license every 3 years. At that time the FCC becomes very interested in how the station has fulfilled its public service requirements. When planning the media campaign, planners can remind the station operators of this fact. They could then promise to write the FCC a letter commending the station for its assistance in promoting the public health and welfare for inclusion in that station's file (as planners of the Stanford study did with much success). This tactic frequently engenders a positive response.

Another important channel factor is coordination (Maccoby and Alexander 1980). Confusion arises when people are exposed to a wide variety of messages from both local and national sources; the sources compete for the attention of the audience and can give contradictory information (Blane and Hewitt 1977). The messages from each medium must be keyed to the others (see Wallack 1978, 1979).

Another important channel factor is targeting the audience that is intended to be reached. The Stanford Heart Disease Prevention Program found that it could not simply translate English-language spots into Spanish to reach California's large Chicano population. Rather, different messages were developed based on the popular “radio-novella” format (Maccoby and Alexander 1980).

Probably the most important factor involving the channel is determining who uses which medium and when. For example, teenagers listen to radio more frequently than they view television (Comstock et al. 1978). Furthermore, FM radio has grown in the last 15 years so that, according to ratings released in August 1979, for the first time it is listened to more than AM radio (Media and Marketing Decisions 1979). In addition, magazines and newspapers that have similar editorial formats also tend to have similar constituencies, and different times of different days attract different audiences to radio and television.

Different markets respond differently as well. For example, television stations in New York City (the Number 1 market) cover a tristate region (New York, New Jersey, and Connecticut), whereas television stations in Los Angeles (the Number 2 market) reach people in only a part of one state. This may be an important consideration if campaigns are mounted on a state-by-state basis.

Yet another channel factor is audience fragmentation. There are, for example, more than 70 radio stations that cover the greater Los Angeles market. There are 14 major formats (disco, soft rock, progressive rock, top 40, jazz, black, country, middle-of-the-road, beautiful music, Spanish, classical, talk, news, and religious), and each station targets its appeal to a different audience. Those who create messages should try to adjust the content and style of radio spots to appeal to the tastes of particular audiences by conforming to the format of the particular station. Anyone who has heard a rock-format spot on a classical music station or a spot featuring Johnny Cash on a black-oriented jazz station can attest to how intrusive the contrast of sounds can be. The incongruity often creates hostility to the message, greatly limiting its ability to persuade. Since adolescents listen to radio more often than they watch television, the potentials as well as the hazards of radio should not be overlooked.

It is essential to recognize the importance of these channel factors in planning a good campaign. Research is needed on the audiences available to particular channels, as is continuous monitoring of the existing knowledge, beliefs, attitudes, risk-related behavior, and media use of those audiences. We should account for the differing level of need and the differences in ethnic group, socioeconomic status, sex (public education messages about alcohol use tend to have a distinctly masculine orientation, according to Blane and Hewitt [1977]), and geographic location (for example, the level of alcohol abuse is much higher in Fresno than it is in Los Angeles).

Receiver Factors

Source, message, and channel can be considered a prelude to what occurs after a person has received a message. Different people respond to the same message differently, often because of personality differences. Studies have shown that some people are more persuadable than others, while certain personal characteristics (i.e., demographic background, emotional stability, ability, motivation, and other personality factors) also play a role in the relative efficacy of a message from one person to another. Although there is quite a literature (see McGuire 1969) on this topic, the discussion in this section is confined to a few pertinent points.

Some factors mentioned before operate as a function of receiver characteristics. For example, in an alcohol campaign the decision whether to mention opposing arguments depends a great deal on the level of intelligence of the receiver, his or her existing predisposition about alcohol, and, most important, the likelihood that she or he will subsequently be exposed to counterarguments. High-fear appeals to those who are prone to anxiety or who have feelings of vulnerability to illness or death should be avoided. On the other hand, they may effectively reach those who feel complacent or invulnerable to the dangers discussed in a campaign. These audiences may be separated by ascertaining patterns of media use through pretesting in target communities.

Self-esteem is also an important factor. McGuire (1969) states that self-esteem creates different reactions to persuasive messages as a function of the relative complexity of the situation. With simple messages, minimal self-esteem is usually sufficient to increase resistance to conflicting information. In more complex situations, however, levels of self-esteem appear to separate levels of effects, probably because when more information is to be processed, a person with lower self-esteem may avoid the message as a means of coping with dissonant information, whereas a person with high self-esteem may not feel as threatened by dissonant information. The optimal persuasion situation would occur with those receivers who have enough self-esteem to actually receive the message and process the information—but not so much that they feel they cannot be told anything new. Unfortunately, this balanced degree of self-esteem and high risk for alcohol abuse are not well correlated.

Age is an important demographic factor in some instances. McGuire (1974) notes that children in junior high school are at an age at which they appear to be most easily persuaded by information contained in a drug (or alcohol) education program. As he notes, this may be fortunate since this age-group is exposed to the greatest amount of initial drug-related information.

In determining the target audience, factors such as self-esteem, intelligence, and persuadability are important, as, of course, are demographic data. Campaign planners should seek to determine the locus of problems of alcohol abuse and the groups who have these problems in common.

Conclusion

Attitudes are complex and require complex strategies for effective change. A message should not try to be all things to all people; it should be targeted to a specific audience, which involves defining who that audience is, what those people are like, what it is we want to change, what the best strategy is to change it, and what is the most efficacious method of disseminating that information. Campaign planners must be willing, in the face of negative evaluative information, to change strategy if necessity warrants. What is most important is change, not the immediate viability of the models.

This approach, by acknowledging that the audience is an equal partner in the communication process, necessitates taking a developmental approach when planning a campaign in order to successfully predict those factors that facilitate acceptance or rejection of messages. Mendelsohn (1973, p. 52) suggests the following conditions for communication programs to be successfully planned and executed: (1) they are planned around the assumption that most of the public to which they will be addressed will be either only mildly interested or not at all interested in what is being communicated; (2) middle-range goals that can be reasonably achieved as a consequence of exposure are set as specific objectives; and (3) careful consideration is given to delineating specific targets in terms of their demographic and psychological attributes, their life-styles, value and belief systems, and mass media habits.

This kind of approach to message design based on principles derived from research was the strategy in the Stanford Heart Disease Prevention Program's 3-Community Study. Butler-Paisley (1975) maintains that future public information campaigns would have a greater likelihood of success if program planners were more cognizant of the social science literature. Program planners and developers should be less intuitive and more empirical in deciding on their creative strategies and more willing to commit the time, money, and resources to weigh alternative strategies.

Caspar Alcohol Education Program

The premise of this program is that specially trained teachers can teach their students about alcohol-related issues rather than alcoholism, modify their attitudes about drinking, and help them develop strong decision-making skills about drinking. It is anticipated that these factors will lead to a reduction of alcohol abuse by teenagers. Peer leaders are used as part of the curriculum to supplement teachers' instruction inside and outside the classroom.

The CASPAR program includes a number of evaluations throughout the course of the intervention, which are used on the assumption that all contingencies cannot be foreseen. Adjustments to the intervention are made on the receipt and evaluation of a variety of surveys.

These studies (1979) have substantiated not only that teacher training is necessary, but also that a properly implemented curriculum is essential for effective alcohol education. The proponents of CASPAR report some change in knowledge and attitudes as well as a gradual change toward moderation of alcohol abuse. While this program shows promise on a number of dimensions, the limited data available at this time preclude an analysis of greater depth.

Alcohol Education Curriculum Project

This prototype is structured on the belief that specially trained teachers, without supplemental peer leaders, can influence attitudes toward drinking that will lead to more responsible decisions by students about their use of alcohol. These teachers are used to teach the curriculum to other teachers. Kits of materials designed for differing grade levels are used.

Evaluations of the curricula on levels of knowledge are made through the use of questionnaires supplemented by self-reported alcohol use by junior and senior high school students. Post hoc analyses have shown a slight influence on teachers' attitudes and that a relationship exists between reported drinking habits and the effect of the curriculum on students. No detailed data were presented to support these contentions, however, and no change in drinking behavior was reported.

University of Massachusetts Demonstration Alcohol Education Project

This project, aimed at college students, attempts to promote a campus environment that is conducive to responsible alcohol use. The premises of the program are:

(1)

Students' drinking behavior influences and is influenced by the environment of the campus.

(2)

This environment can be changed by extended efforts conducted simultaneously throughout the social system.

(3)

The use of alcohol by students can be influenced by using a combination of mass media to promote awareness (setting the agenda) and small-group interactive approaches to examine alcohol-related attitudes and behavior.

The goals of the program are to increase the knowledge of those in the university community about responsible alcohol use, to help individuals understand and moderate personal and social alcohol-related behaviors, and to promote early identification of those with alcohol-related problems. The program uses a comprehensive set of specific approaches including media campaigns, peer-leader training, seminars, in-service training of service staffs, and attention to institutional policies to facilitate early identification and treatment of those with alcohol problems. Evaluation data include mailed attitude surveys, random monitoring of dormitory behavior, interviews with dormitory directors, and data on alcohol-related arrests, on-campus sales, medical incidents, and property damage.

Discussion

Summary reports (1979) of these projects indicate that there has been some change in attitudes about alcohol use, but the reports present no specific data. The ratio of alcohol-related behavior to total behavior (neither is defined) in dormitories (Amherst, Massachusetts, project) evidently decreased from spring 1977 to fall 1977, but insufficient data are supplied with which to make a proper evaluation of this finding.

It would appear that this program holds a great deal of promise in principle, since a well-integrated mass-media and group-counseling approach is used. The major problems at this time appear to be that the questionable reliability and validity of the data hinder inferences about alcohol-related behavior change; the program has not been implemented for a sufficient amount of time for a meaningful assessment of behavior change; and the absence of a control group design makes it difficult to interpret the significance of the data that are reported.

Those who wish to mount campaigns with the ultimate goal of changing behavior must be willing to invest more than 6 months before deciding the success or failure of their efforts and to stay cognizant of the quasi-experimental nature of their efforts—and hence the need for use of quasi-experimental design features as development of controls. As the Stanford Heart Disease Prevention Program's 3-Community Study showed, unlearning old behaviors (which may have taken years to develop) by replacing them with new ones is a process that will, in all likelihood, take a long period of time and require careful design in order to demonstrate effects.

In brief, on the basis of general principles, the two projects from Massachusetts show the most promise. For example, student peers make credible, trustworthy sources to disseminate messages and to lead group discussions. The University of Massachusetts program uses small-group approaches led by trained peers to provide positive social situations for discussing alcohol and behavior. This combines the use of trustworthy, credible peers with the beneficial effects of the open participation of people in small, interactive groups. The CASPAR program also makes use of peers to disseminate information and lead group discussions.

They differ greatly from the formal classroom situation (used in King County), which does not lend itself to frank discussions, peer assistance, or meaningful problem solving. Teachers may not be seen by students to be trustworthy sources of information, since an adversary relationship frequently exists between students and faculty. Furthermore, the reliance on a self-reports of alcohol use administered by teachers may be seen by students as a tactic to catch them engaging in a proscribed activity. This may heighten the lack of trust, which will not help to modify the problem.

The reliance of the University of Massachusetts program on FM radio indicates the recognition by program planners that the use of television for information and entertainment is not ubiquitous. Rather, college students tend to rely more on radio than television (Comstock et al. 1978) than does the average person. In a college community that is not within a major metropolitan area (such as Amherst, Massachusetts), student participation in as well as the use of college radio may be quite substantial. This may be true of similar campus communities; it can easily be ascertained through a preintervention survey.

There is insufficient evidence in NIAAA's reports of how closely these programs adhere to all the principles described in this paper. That would require minute descriptions of, for example, the procedures by which the usefulness of sources used was determined vis-à-vis their availability to program planners; the subjects' media use vis-à-vis the funds, facilities, and expertise available; and the usefulness of the programs versus the nature and extent of the target population's alcohol use.

Rather than serve as a scorecard for NIAAA or other programs, this paper should serve as a guide to those implementing these and future programs in the creation of present and future interventions, taking local factors into consideration.