Much of the research on adolescent drug use (and other areas of crime and delinquency research) concerns itself with the simple relationships between variables which are often demographic in nature (for example, ethnicity, age, socio-economic status, political leanings, family make-up, etc.) and drug use, again without attempting to explain the relationships in further detail. For example, Erich Goode's (1970) book "The Marijuana Users" describes the social and demographic characteristics of the marijuana users in his study as typically being young, male, mid- to high-class, college educated, more likely to be politically liberal or radical, more sexually permissive, and less authoritarian. While these demographic factors tell us something about the background and social position of the majority of marijuana users, they fail to explain why someone fitting these characteristics might, or even might not, use marijuana. What is missing is the explanation as to why adolescents (or adults) use marijuana, how they acquire the skills to use marijuana for the purposes of getting 'high', how they do so in the face of negative sanctions from much of society and the law, and how these factors are transmitted to the individual. In short, the question posed here is not "what are the characteristics of marijuana users" but "how does one become a marijuana user?"

To answer this question, I will begin with a discussion of early studies of marijuana use, and then move on to explore Akers' (1985) social learning theory as representative of the types of theories which more closely explore issues of delinquency by investigating the behavioural processes through which delinquent activity is acquired.

Marijuana Use as A Social, and a Learning Process -- Early Works

Possibly the most influential study on marijuana smoking is Howard Becker's (1953) study "Becoming a Marihuana User." The focus of the study was to identify common elements of "changes in attitude and experience which lead to the use of marihuana for pleasure" among his sample of 50 male marijuana users (p. 235; italics in original). An important focus of the study was also to identify the requirements not only for initial use, but of continued use (whether intermittent or more continuously). He identified three main processes in learning to use marijuana for pleasure.

First, the user must learn to smoke marijuana in a way which produces the pharmacological effect of the drug. For example, marijuana smoking typically involves a process different from that of tobacco smoking. The subjects in Becker's study indicated that a novice marijuana smoker does not normally get 'high', simply because he does not inhale the smoke properly; that is, hold it in the lungs for a period of time, and then exhale. As Becker observed: "No person continued marihuana use for pleasure without learning a technique that supplied sufficient dosage for the effects of the drug to appear" (p. 237). Becker further notes that this process need not be learned solely through direct teaching of the technique. For example, a new user ashamed to admit that they do not know how to smoke marijuana may try to learn it through observation and imitation (p. 237).

Second, learning how to obtain an effect from the drug must be followed by learning to recognize the effects of the drug. If a user is to continue use, he or she must learn what the effects of the drug are. Again, this learning of the effects of the drug can be direct (in the case of the user being told what to look for) or indirect (in the case of the user looking for cues from other users as to what the effects of the drug are).

Third, in order for the user to continue the use of marijuana, he must learn to enjoy the effects of marijuana. Becker states that the effects of marijuana are not "automatically or necessarily pleasurable" (p. 239), and the user must acquire a 'taste' for the effects of the drug. This is accomplished through a redefinition of the effects of the drug as pleasurable for those who found their first experience unpleasant, and the definition of the drug's effects as being pleasurable for others who had an indifferent or pleasurable first experience. Without this definition or redefinition, Becker asserts, use will simply not continue.

Becker (1963) continues this line of investigation in Chapter 4 of Outsiders, in which he discusses the social controls which must be neutralized for a marijuana user to continue use. The focus of the study is to answer the question: "What is the sequence of events and experiences by which a person comes to be able to carry on the use of marihuana, in spite of the elaborate social controls functioning to prevent such behavior?" (p. 60). The social controls the user must neutralize in order to initiate or continue use are limited supply of the drug, the need for secrecy, and the public definitions of marijuana use as immoral. Becker analyzes the role of these three controls in relation to beginning, occasional, and regular use of marijuana.

First, all users must find a supply of marijuana, and typically do this through participation in marijuana using groups in which the supply of the drug is occasionally or more continually found. Becker (1963) states that a precondition for moving to a more regular level of use is precipitated by changes in group participation and membership which lead to changes in the user's level of use by affecting the individual's access to marijuana (p. 66). In short, the control over supply of the drug is not through direct legal sanctions, but by legal sanctions which render the source of the drug unreliable (p. 66). The way around limitations of supply are through social relationships and membership in groups which do have access to marijuana, and through the user becoming sensitive to the "controls of the drug using group, so that there are forces pressing toward use of the new sources of supply" (p. 66).

Additionally, in order to move to a higher level of use, the user must deal with the social controls which make secrecy a defining feature of marijuana use. Secrecy is important to limit the probability that the user will be caught using marijuana, and punished either through legal or more informal social prohibitions (such as family members finding out about one's use). For beginners, Becker writes, the overcoming of the threat of sanctions is especially important (and, therefore, so are the steps taken to be secretive). However, each successive level of use (from beginner to regular user) must be preceded by a realization that secrecy can be carried out fairly easily, and thus the fear of being caught by nonusers is limited by this realization. Becker here postulates a negative correlation -- that marijuana use is limited in proportion to the fear that others will find out about their marijuana use and react in some "punishing" way (p. 72). As the level of fear decreases, the level of use is likely to increase, and this can take place only when the user revises his conception of the dangers involved (p. 72).

Becker's third example of social control, conventional notions of morality, need to be neutralized or rationalized in order for the user to begin or continue use. These conventional notions of morality are informed by the common stereotype of the marijuana user as a person incapable of being responsible for his own welfare, and incapable of rational behaviour (pp. 72-73). Becker illustrates this stereotype with the use of a quote by H.J. Anslinger, who was instrumental in bringing about marijuana prohibition in the United States. However, a more extreme example can be found earlier in Canadian Justice Emily Murphy's (1922/1973) anti-drug book The Black Candle, which was instrumental in the placing of marijuana in the Criminal Code in 1923, a year after this book was published, and almost ten years before anyone was charged with marijuana possession in Canada (p. 332):
 

Charles A. Jones, the Chief of Police for the city [of Los Angeles], said in a recent letter that hashish, or Indian hemp, grows wild in Mexico but to raise this shrub in California constitutes a violation of the State Narcotic law. He says, "Persons using this narcotic, smoke the dried leaves of the plant, which has the effect of driving them completely insane. The addict loses all sense of moral responsibility. Addicts to this drug, while under its influence, are immune to pain, and could be seriously injured without having any realization of their condition. While in this condition they become raving maniacs and are liable to kill or indulge in any form of violence to other persons, using the most savage methods of cruelty without, as said before, any sense of moral responsibility ... If this drug is indulged in to any great extent, it ends in the untimely death of its addict."

Obviously, the moral tone of this passage is unmistakable. Of particular noteworthiness are the statements regarding the marijuana user's inability to be responsible for his own welfare and behaviour. Becker indicates that the beginning user would quite possibly hold some of these views, and in order to initiate marijuana use, would have to "accept an alternate view" (p. 73) and find ways of neutralization these stereotypical conceptions of marijuana users. These rationalizations, Becker continues, are to be found in the marijuana using groups themselves. Becker summarizes (p. 78):

In short, a person will feel free to use marijuana to the degree that he comes to regard conventional conceptions of it as the uninformed views of outsiders and replaces those conceptions with the "inside" view he has acquired through his experience with the drug in the company of other users.

Erich Goode's (1970) book The Marijuana Smokers can be seen (for the purposes of this paper) as both a replication and extension of Becker's early work. In discussing the extent to which marijuana use had spread in the late 60s United States, Goode focuses on the fact that, at the time, the use of marijuana was largely influenced by pro-marijuana messages which were much more common than anti-marijuana messages, the latter being negated by the fact that many critics of marijuana had never used the drug. Thus, to adolescents who were much more experiential than marijuana's detractors, the anti-marijuana message carried less weight in American society.

Focusing on the micro effects of the macro environment in which marijuana use took place, Goode summarizes the dominant role of five factors in this process (p. 127):
 

• The initiate's perception of danger (or lack thereof) in marijuana use

• His perception of its benefits

• His attitude toward users

• His closeness to marijuana's endorsers

• His closeness to the individual trying to turn him on

What remains important for this paper, however, is the quality or character of the factors Goode has distilled in his study. As with Becker's two papers on the factors involved in marijuana use, these factors involve a) a process through which a person comes to use marijuana, b) a conception of the risks and benefits involved, c) a definition of use as being good or bad, and d) the fact that learning to use marijuana is very much a social phenomenon. It is these main factors which are also shared in other theories of marijuana use, the most pertinent (and, the subject of this paper) being social learning theory (SLT). It is to the discussion of SLT I now turn.

Enter Social Learning Theory

Social learning theory, as described by Akers, Krohn, Lanza-Kaduce and Radosevich (1979), is an attempt to incorporate differential association theory and modern behaviourism into sociological theory (p. 637). It takes the form of an amalgamation of Sutherland and Cressey's (1969) differential association theory, and incorporates behavioural reinforcement theory into a theory which is applied to the prediction of deviant behaviour. In contrast, Akers et al point out that other theories focus almost primarily on structural variables (for example, class, race, anomic conditions, breakdown in social control) and how they affect delinquency rates. SLT, instead, stresses behavioural mechanisms by which these variables produce the rates. For this reason, Akers et al assert, SLT is quite complementary to other theories of deviance, and can be used to form more complex models of delinquency through its integration with other theories (p. 637).

Differential Association Theory

The first component of SLT, differential association theory, has remained largely unchanged over its history. Differential association theory assumes that criminal acts occur when a situation appropriate for it, as defined by the person, is present (Sutherland and Cressey, 1969, p. 80). Differential association theory is presented as a series of 9 statements, which are summarized here (while keeping much of the original language) according to three general areas: 1) the learning component, 2) the differential association component, and 3) a statement of what criminal activity is not (Sutherland and Cressey, 1969, pp. 80-82; formatting mine):

Criminal behaviour is learned in interaction with other persons (typified as intimate, personal groups) in a process of communication. Further, the learning includes a) techniques of committing the crime and b) the specific direction of motives, drives, rationalizations and attitudes. Finally, the process of learning involves all of the mechanisms that are involved in any other learning.

The specific direction of motives and drives is learned from definitions of the legal codes as favorable or unfavorable. Most importantly, a person becomes delinquent because of an excess of definitions favourable to violation of law over definitions unfavourable to violation of law. These differential associations may vary in frequency, duration, priority and intensity.

While criminal behaviour is an expression of general needs and values, it is not explained by those general needs and values, since noncriminal behaviour is an expression of the same needs and values.

The "heart" of differential association theory, according to Akers (1985), is that "one commits criminal acts because his accepted 'definitions' of law as something to violate are in 'excess' of his accepted definitions of the law as something that can, must, or should be obeyed" (p. 40; italics in original). Additionally, these definitions (and any resulting delinquency) are not simply the result of one having 'bad' friends, but depend on the balance of 'good' to 'bad' definitions of the law (p. 40).

Operant Conditioning

The other important aspect of social learning theory, operant conditioning, is derived here from Akers, et al's discussion (1979, pp. 637-639). Social learning theory's primary learning mechanism is operant (or instrumental) conditioning, in which behaviour is shaped by the stimuli which follow as a consequence of that behaviour. Under the rubric of operant conditioning, social behaviour is acquired through direct conditioning or through the imitation or modeling of others' behaviour.

In operant conditioning, a person's actions may be positively reinforced (for example, there is a reward, social or nonsocial, which is applied as a consequence of the action), or negatively reinforced (in which the consequences of that action would be seen as punishment, but are avoided). Alternately, behaviour may be weakened through positive punishment (a stimuli which is aversive to the individual) or through negative punishment (where there is a loss of a reward for that action).

The result of these reinforcements is that the continuance of behaviour is dependent on past and present rewards or punishments for a certain behaviour, and also rewards or punishment attached to behaviours that may arise from alternate behaviour. For example, using this model, deviant behaviour (and the continuance of deviant behaviour) may be seen to be contingent upon both the rewards or punishment arising from the deviant behaviour, as well as the rewards or punishment arising from alternate, non-deviant behaviour. Central (and most important) to this explanation is that a person would assess the rewards or punishment involved in a certain act, and the alternate of doing that act. Akers et al call this Differential Reinforcement.

The Synthesis of DAT and OC

In the empirical testing of social learning theory as it applies to adolescent marijuana and alcohol use, Akers et al (1979) distinguish four main factors which embody the above concepts. The first two, Differential Association and Definitions, are each related and are derived primarily from differential association theory. Here, the process through which definitions are learned is through differential association with significant groups. While Sutherland and Cressey (1969) describe significant groups as being primarily significant peer-friendship and family groups, Akers et al also concede that these primary groups with which one is in differential association with can be schools and churches and other groups.

Occupying both differential association theory (which, of course, includes a learning component) and operant conditioning, is Imitation. This factor is included in Akers et al's empirical test of adolescent marijuana and alcohol use primarily because it is an indication of one of the primary ways in which behaviour is learned (along with direct conditioning), and is theoretically applicable to adolescents in the study who are new to using either of these drugs.

Finally, the concept of Differential Reinforcement (extracted from operant conditioning, but cast in differential association terms) is used. The concept of differential reinforcement is further subdivided into social and social/non-social reinforcement. The former is comprised of factors indicating what the possible reactions of family and friends would be, encouragement not to use (from parents and friends), and informal and formal deterrence. The latter is comprised of the overall social and non-social rewards minus the costs of using, overall reinforcement balance (a subjective assessment of whether mostly good or mostly bad things would happen), and the usual effects of the drugs when used (p. 655).

Finally, Akers et al propose a causal ordering to these variables. First, differential association with peer-friendship or other groups provides definitions, imitation of models and social reinforcement (in this case, of use or abstinence). Next, definitions are learned through imitation and social reinforcement by the group. These definitions produce the inital use or continued abstinence. Finally, actual consequences (social and non-social) determine the probability the behaviour will continue (p. 638).

Akers et al summarize their theory this way: "the youngster associates with peers who are users, learns definitions favorable to use of the substance, and then uses (p. 639-640).

The Empirical Tests

Akers et al's (1979) paper, "Social Learning and Deviant Behavior: A specific test of a general theory" provided the first impressive glimpse of the power of social learning theory to predict adolescent marijuana use. The sample for the study was approximately 2,400 children and teens in grades 7 through twelve across three mid-western states in the U.S. The dependent variables were alcohol and marijuana use and abuse, while the independent variables were the social learning theory concepts of differential association, imitation, differential reinforcement, and definitions. Quite simply, the results showed that all independent variables were strongly related to marijuana use and abuse. The full SLT model explained 68% of the variance in marijuana use, and almost 39% of the variance in marijuana abuse. The strongest predictor of marijuana use and abuse was differential association, while the weakest was imitation, which in the case of marijuana abuse explained almost none of the variance. (As indicated earlier, imitation is especially important at the early or pre-use stages of substance use, so the low predictive value of imitation in marijuana abuse is not surprising.)

In two other articles, Akers and others (Lanza-Kaduce, Akers, Krohn, and Radosevich, 1984; and Akers and Cochran, 1985) have used the same data set to look at whether SLT could discriminate between continued users and ex-users of marijuana and other drugs, and to make a comparison of the predictive powers of SLT, social bonding theory, and anomie (strain) theories on adolescent marijuana use. In the first article (Lanza-Kaduce, et al, 1984), SLT theory concepts were applied using a discriminant analysis on continued users and ex-users of 5 drugs, which included alcohol, marijuana, stimulants, depressants, and 'stronger drugs'. The most discriminating concept of SLT was differential association, while the least discriminating was imitation and the non-social differential reinforcement (as indicated by the 'usual effects' of the drugs). In the second article (Akers and Cochran, 1985), SLT explained 68%, social bonding theory explained 30%, and strain or anomie theory explained 3% of adolescent marijuana use, clearly indicating that social learning theory was better equipped to predict adolescents' use of marijuana than were social bonding or strain theories.

In other studies, social learning theory has again proved valuable in predicting adolescent marijuana use. For example, in Winfree and Griffith's (1983) study exploring rural adolescents' marijuana use in 1975 and 1979, SLT explained 34% of the variance in adolescent marijuana use in the 1975 sample, and 52% of the variance in the 1979 sample. Of importance in this study, however, was the addition of certain variables (such as perception of severity of drug laws and peer pressures to use drugs) which were analyzed as part of the SLT model. The results gave strong support to Winfree and Griffith's contention that the 1975 and 1979 samples viewed their world quite a bit differently in terms of drug-related attitudes and behaviors, since the 1979 respondents were attitudinally more conservative, less inclined to agree with their peers over drug-related issues, yet a higher percentage of them had used marijuana.

One important question to be asked here is why do these studies so convincingly support social learning theory? One answer is that marijuana use, particularly adolescent marijuana use, is superbly suited to social learning theory. For the explanation of why this might be, it is instructive to re-examine some of the assertions of Becker and Goode.

The Suitability of Social Learning Theory in Exploring Adolescent Marijuana Use

Winfree and Griffiths (1983) note that Becker's observations that marijuana use must be learned and that even the pleasurable effects of marijuana must be recognized went largely unheeded by many drug researchers for many years. It is clear to Winfree and Griffiths that social learning theory supplies the necessary causal ordering and "conceptual parsimony" (p. 220) that had been missing in the application of other theories of deviance to the question of drug use. While social learning theory might draw heavily from operant conditioning and differential association theories, it also shares many commonalities with Becker's (1953, 1963) and Goode's (1970) statements on marijuana use. I will begin this section by summarizing Akers et al's (1979) statement of the causal ordering of social learning theory variables, and then detail the similarities between SLT and Becker and Goode's work.

According to Akers et al, the first step in the learning process occurs with differential association; that is, the "interaction and identity with different groups" (p. 638). Through differential association, a social environment is enabled which makes it possible for an adolescent to be exposed to definitions, learn through imitation of models, and receive social reinforcement of activity (in this case, reinforcement for the use or abstinence of different substances). The learning of definitions takes place through imitation, and the social reinforcement of the definitions by other members of the group. These definitions are granted a high importance in social learning theory, since they may be learned through imitation, gain strength through social reinforcement, and can also function as cues for the individual to use or remain abstinent.

In the initial stages (for example, when the individual is confronted with the opportunity to use), definitions are often learned through imitation, and combined with anticipated social and/or non-social rewards, produces either the initial use or continued abstinence. Imitation at later stages, however, becomes less important since the definitions may be influenced by more direct factors -- the actual social and/or nonsocial reinforcers or punishers resulting from use or continued abstinence may modify the definitions. Of course, they might not modify them in any appreciable degree, however, this could be seen as a strengthening of the definitions through their continued existence in the same form. Finally, the actual social and/or nonsocial reinforcers or punishers could take place at the time of the use or abstinence, as in the case of the reaction of a peer group, or the actual effect of the substance itself, or in the future, as in the case of the reaction of family members to the use (or non-use).

Becker, Goode and Social Learning Theory

It has already been stated that Becker (1953) has specified that the user must learn to use marijuana properly, learn to distinguish its effects, and then learn to enjoy those effects if the user is to continue marijuana use. In this simple statement, much of social learning theory is relevant. First, while marijuana use might be accomplished by a single person, it is exceedingly unlikely that there might be individuals who procured, learned to use, and learned to enjoy the effects of marijuana while not in the presence of others. Thus, differential association will usually take place, if not the very first time, then later on, and certainly before a person could learn to use the drug, learn to distinguish its effects, and learn to enjoy those effects. Becker's (1963) treatment of the social control of limited supply also is relevant here, since Becker states that the first social control to be overcome if one is to progress to use is the limited supply. Typically, this involves participation in marijuana using groups (differential association), through which a supply of marijuana may be found.

One criticism of both social learning theory and differential association is that they specify that the deviant or delinquent behaviour must be learned. In some forms of deviance (for example, stealing, lying, etc.) one can commit deviant acts without having been taught that act. In the case of adolescent marijuana use, however, it is clear that the use of marijuana could not ideally be performed without some prior knowledge of the act. Even if the learning comes from imitation of television or movie characters, and the supply of marijuana is found by the individual, there is still both a learning process and a social (albeit one-sided) event happening which may be enough to supply definitions necessary to the committing of the act. However, learning the effects of the drug, and determining that the effects are pleasant will probably not occur.

Once differential association takes place, the individual is exposed to definitions, learns through imitation, and receives group reinforcement or punishment for his or her actions. The group, whether pro-marijuana or anti-marijuana, will determine what type of definitions are supplied to the user, and also determine what punishers or reinforcers it is capable of delivering to the individual. In the case of a pro-marijuana group, the definitions supplied will be of a type which would make marijuana use more likely, unless there is a source of definitions available to the individual which prohibit the use of marijuana, and are of sufficient strength to counter the definitions supplied by the pro-marijuana group. These anti-use definitions can arise from differential association with anti-use groups, and also from prevailing societal notions of morality, as Becker (1963) outlines as the third social control to be neutralized in order for use to occur. In the initial stages of the individual's use (or non-use), the balance of definitions favourable to use as opposed to abstinence is critical.

In Goode's (1970) 5-factor formulation of the conditions involved in use, the initiate's perception of danger is important in determining whether use will occur. Here, definitions supplied by the group, and learned through imitation, combined with the anticipated rewards or punishments for using (of which danger would be a factor) will contribute largely to whether or not use will occur. Additionally, the new user might learn through observation of the group members that there is no danger of certain precautions are taken. This dovetails with Becker's (1963) second social control that the marijuana user must negate: secrecy. Secrecy, Becker writes, is important to limit the possibility and fear of being found out or caught. However, one's fearfulness of being caught (a definition or marijuana use as risky) can be balanced by, or eclipsed by learned definitions from a pro-marijuana group, who are less fearful of being caught, and know techniques necessary to limit their chances of being caught. These definitions are communicated to that individual either through his or her direct learning, or vicariously through imitation.

Goode (1970) also indicates that important factors which determine use are the perceptions of the benefits of use, the attitude toward users, the closeness to marijuana's endorsers, and the closeness to the person trying to 'turn on' the individual. Again, whether at initial use or later on in the career of the marijuana user, differential associations and definitions remain extremely important. The adolescent's perceptions of benefits of using marijuana are definitions supplied by the group, whether directly or whether the adolescent learns them through imitation. The adolescent may also learn them through observation of the reinforcement of punishment of the group members' actions in using marijuana, whether social (due to censure by others) or non-social (through observation of the group members' enjoyment of the experience).

Goode's factor of the individual's attitude towards users may be shaped in much the same way, although certainly an important factor for the beginning user is the attitude towards users (i.e. the definition of use and users) supplied to him or her by other groups, and by the type of conventional notions of morality Becker (1963) describes. Since definitions are learned, and can change over time and especially with further involvement with the marijuana using group, attitude towards users may be affected by closer involvement with the marijuana-endorsing group. When definitions not favourable to marijuana use are outweighed by definitions favourable to use, and when definitions of the marijuana using group as immoral, etc., are outweighed by definitions favourable to the group, use is more likely to occur. Both Goode's factors of closeness to the marijuana-endorsing group and the closeness of the individual to the person trying to 'turn him on', are therefore subsumed under social learning theory.

Social learning theory has a cyclical nature as well, in that differential association can lead to definitions favourable to an act, which leads to the act itself. At this point, rewards or punishments are fed back into the definitions favouring (or, not favouring) a certain act. Becker's (1953) description of the types of learning that must be accomplished before one becomes a use of marijuana also contain the same kind of causal loop, in that continued use requires that each 'lesson' be learned in succession. For example, once the user has learned how to obtain an effect from the drug, he or she must look for cues as to what those effects are through observation or imitation. This is exactly what social learning theory predicts, in that the individual must learn from the observation of others, and then redefine the conception of the effects of the drug as having certain characteristics. This involves, in social learning theory language, the use of marijuana followed by the differential reinforcement of the experience, which, combined with imitation of others, modifies definitions of the effects of the drug. Becker (1963) continues this idea through to the learning of the effects of the drug as being pleasurable. In social learning theory, this can be seen as another feedback of differential reinforcement into modified definitions of marijuana use as pleasurable. In either theory, if use cannot be defined (or redefined) as pleasurable, use is not likely to continue.

To sum, then, social learning theory can be seen to incorporate all of the important aspects of Becker's (1953, 1963) and Goode's (1970) work. What is interesting here is not so much that social learning theory incorporates most of Becker's and Goode's observations, but that the main areas of commonality of all three, specifically differential association and definitions, supply most of the variance in adolescent marijuana use in the studies outlined above. They are also able to discriminate more than other social learning factors between ex-users and continued users of marijuana in the Lanza-Kaduce et al (1984) study.

It should also be noted that in a review of literature on adolescent marijuana use, social learning theory appears to be the best-fitting and more empirically justified model in this area. Again, while this might be a reflection of the data used (recall that three of the social learning studies mentioned above utilized the same data set), more likely this is because of the applicability of this theory to adolescent marijuana use, given that it is a) a social phenomenon (which makes differential association a necessary part), b) an activity where one's definitions are quite possibly more polarized and less homogenous than in other topics, and c) and activity which is learned. Had any one of these conditions not been met in the case of adolescent marijuana use, it is unlikely that social learning theory would display as great a predictive power as it currently does.

Other Issues and Further Research

While social learning theory shows much promise, at least in terms of adolescent substance use, several issues that are central to most theories of delinquency need to be discussed briefly.

Age, gender and ethnicity have proven problematic for most theories of delinquency. For example, many theories do not attach great importance to whether or not female delinquency is explained (for example, Hirschi, 1969). As well, a theory which concentrates on factors leading to deviance many have trouble determining why delinquency typically peaks in the late teens and then declines as the teen enters into adulthood, if the theory doesn't take into account change in these factors (for example, Hirschi and Gottfredson, 1990). Finally, a theory of delinquency or deviance should be able to be meaningful when applied to different ethnicities and cultures.

With respect to adolescent substance use, social learning theory has performed well in the areas of gender and age, and less well in the case of ethnicity. For example, White, Johnson and Horwitz (1986) found that differential association variables predicted adolescent substance use almost equally well among both genders and three age groups (p. 362). As well, Dembo, Grandon, La Voie, Schmeidler and Burgos (1986) noted that social learning theory can predict the role parents and parents drug use can have in adolescent substance use, and that this prediction was not affected by ethnicity or gender (p. 85). Finally, Marcos and Johnson (1988) also used social learning theory to predict adolescent drug use in two environments: Greece and the United States. However, they found limited support for social learning theory in Greece, in contrast to the predictive value of the theory when applied to adolescents in the United States. Their findings are weakened by the fact that they had a small sample in Greece (174 compared to 2,610 in the U.S.), and the fact that their model is basically a control theory model, with the addition of only a single variable encompassing differential association.

A theoretical problem exists with almost all theories of delinquency, and that is what the source, or impetus, of motivation for an individual to commit a delinquent act is. Strain theory handles this directly by hypothesizing that an individual with needs that are not satisfied by society will likely commit a delinquent act to have those needs met. In other words, it is the strain in not being able to legitimately fulfill personal needs which causes delinquency (Akers and Cochran, 1985).

Hirschi's (1969) social control theory does not deal with motivation much more than to indicate that the specific source of motivation does not matter much; all that is required is that it is present, in whatever form. In seemingly much the same way, social learning theory is also decidedly silent on the topic of motivation. In hypothesizing that adolescents will learn, through differential association, definitions that will influence their behaviour, social learning theorists do not have to say anything explicit about motivation to crime.

However, to return to the subject of differential association, it can be seen that what is important for DAT in terms of motivation is that the of the individual is directed: "The specific direction of motives and drives is learned from definitions of the legal codes as favorable or unfavorable" (Sutherland and Cressey, 1969, p. 81). For social learning theory, the explanation must be seen to be much the same, in that since motivation is not mentioned, it is assumed that motivation to behave (regardless of the way one behaves) is present, and that differential association, definitions, and differential reinforcement serve to direct that motivation either towards a specific behaviour or its alternate.

Perhaps the resolution of the topic motivation can be best found in Becker's (1953) observation that the use of marijuana is inextricably linked to the learning process one must go through in order to use marijuana. In attempting to discount 'trait' theories of delinquency (in which a presupposed trait is responsible for the motivation to commit a delinquent act), Becker writes (p. 235):

The motivation or disposition to engage in the activity is built up in the course of learning to engage in it and does not antedate this learning process. For such a view it is not necessary to identify those "traits" which "cause" the behavior. Instead, the problem becomes one of describing the set of changes in the person's conception of the activity and of the experience it provides for him.

This explanation is entirely consistent with social learning theory's proposed causal chain of differential association, definitions, and behaviour. In locating motivation in the learning of definitions, and by adding to that the idea that an individual's behaviour is dependent upon the definitions attributed to that behaviour, and the anticipated real or imagined rewards or punishments that will follow that behaviour, it can be seen that motivation towards one behaviour or another can be located in the learning process itself. For example, regardless of the activity, if the ratio of definitions are on the side of being favourable to the behaviour, and there are few if any realistic punishments for that behaviour, we can predict that the behaviour will occur. Perhaps, in the final analysis, perceived or real reinforcements and rewards of the behaviour (whether social or not) are enough to motivate the individual in the direction of the behaviour. On the obverse, perceived or real punishments and sanctions would seem to be enough to predict that an alternate behaviour will be chosen. In any case, it is reasonably clear that an adolescent's differential association with a marijuana-using and promoting peer group might serve to motivate that adolescent in the direction of using if he or she faces that decision, since that decision carries with it the promise of closer ties with that group if he or she uses, and punishment (even if only indifference) from the group is he or she doesn't use.

Typically, research using social learning theory has used cross-sectional data (Akers, Krohn, Lanza-Kaduce and Radosevich, 1979; Lanza-Kaduce, Akers, Krohn and Radosevich, 1985; Akers and Cochran, 1985) or a cohort approach (Winfree and Griffiths, 1983). However, there is a large element of causal priority given to the different concepts of social learning theory which call out for a closer look at whether or not the causal ordering given to the elements in social learning theory are in fact as they are hypothesized.

Much of the research in social learning theory specifies the same causal path that Akers et al (1979) specify, but focus almost primarily on the use of ordinary least squares regression to test the model as a whole. While support for the model has been found, OLS regression also carries with it no small problem in being able to specify causal direction among independent variables. In the question of adolescent peer groups and individual delinquency, social learning theory hypothesizes that differential association precedes individual behaviour (Akers et al, 1979). In contrast, Hirschi's (1969) social control theory hypothesizes that individuals commit delinquent acts before a peer group is found -- Hirschi refers to this as the 'birds of a feather' effect.

In the case of adolescent marijuana use, it can be safely said that the peer group is instrumental in teaching the individual the necessary skills to use marijuana, which social learning theory predicts. However, much less can be said about the empirical testing of other causal links in social learning theory. While social learning theory states an explicit causal chain, and that chain happens to be logically sound, the issue of a causal loop (described previously) is less clear. For example, how is a study to be designed which can actually test ideas of differential association leading to definitions, leading to behaviour, leading to reinforcement, and finally leading back to definitions to begin the process again? How is each loop to be differentiated according to causal ordering? How would it be possible to determine, temporally, where an adolescent happens to be in this causal loop? Unfortunately, these questions cannot be answered here, but must be considered further, especially in cases of delinquency which are not as 'tidy' as adolescent marijuana use, and which therefore might contribute less support for social learning theory.

Conclusion

Social learning theory obviously offers a great deal of explanatory power in issues of adolescent drug use. It has shown also that is it quite compliant with previous theories of marijuana use, and is derived from much of the same factors and characteristics. It also explains to a significant degree all aspects of marijuana use, from pre-use to beginning use, from occasional use to continued use, right through to cessation.

However, much of this predictive power is derived from the unique applicability of social learning theory to adolescent marijuana use. Several aspects of the theory, including the general finding that marijuana use is learned, and further, that marijuana use is learned through association with peer-friendship groups, make social learning theory an important part of the literature on the etiology of drug use, which itself has suffered from too tight a focus on demographic and other social variables, and a lack of latitude outside the domain of rates of use.

While social learning theory fills a need in the drug use literature, several unresolved problems remain. Most importantly for types of deviance other than marijuana use, the causal ordering of the theory, along with its causal loop, must be further described and empirically tested. As well, more research needs to be undertaken to try to apply this theory cross-culturally, and to determine if social learning theory is simply an American theory of social learning, or a human theory of social learning.

Bibliography

Akers, Ronald L., Marvin D. Krohn, Lonn Lanza-Kaduce, and Marcia Radosevich (1979) "Social Learning and Deviant Behavior: A specific test of a general theory." American Sociological Review, 44, August, 636-655.

Akers, Ronald L. (1985) Deviant Behavior: A Social Learning Approach. 3^rd Edition. Belmont, California: Wadsworth Publishing Company.

Akers, Ronald L., and John K. Cochran (1985) "Adolescent Marijuana Use: A test of three theories of deviant behavior." Deviant Behavior, 6, pp. 323-346.

Becker, Howard S. (1953) "Becoming a Marihuana User." The American Journal of Sociology, LIX, 3, pp. 235-242.

Becker, Howard S. (1963) Outsiders: Studies in the Sociology of Deviance. London: The Free Press of Glencoe.

Dembo, Richard, Gary Grandon, Lawrence La Voie, James Schmeidler and William Burgos (1986) "Parents and Drugs Revisited: Some further evidence in support of social learning theory." Criminology, 24, 1, pp. 85-104.

Hirschi, Travis (1969) Causes of Delinquency. Berkeley: University of California Press.

Hirschi, Travis and Michael R. Gottfredson (1990) A General Theory of Crime. Stanford: Stanford University Press.

Hundleby, John D., and G. William Mercer (1987) "Family and Friends as Social Environments and Their Relationship to Young Adolescents' Use of Alcohol, Tobacco and Marijuana." Journal of Marriage and the Family, 49, February, pp. 151-164.

Goode, Erich (1970) The Marijuana Smokers. New York: Basic Books.

Lanza-Kaduce, Lonn, Ronald L. Akers, Marvin D. Krohn, and Marcia Radosevich (1984) "Cessation of Alcohol and Drug Use Among Adolescents: A Social Learning Model." Deviant Behavior, 5, pp. 79-96.

Marcos, Anastasios C., and Richard E. Johnson (1988) "Cultural Patterns and Causal Processes in Adolescent Drug Use: The Case of Greeks Versus Americans." The International Journal of the Addictions, 23, 6, pp. 545-572.

Murphy, Emily (1922/1973) The Black Candle. Toronto: Coles Publishing Company.

Sutherland, E., and D. Cressey (1969) Criminology. 10^th Edition. New York: Lippincott.

Warr, Mark and Mark Stafford (1991) "The Influence of Delinquent Peers: What they think or what they do?" Criminology, 29, 4, pp. 851-866.

White, Helene Raskin, Valerie Johnson and Allan Horowitz (1986) "An Application of Three Deviance Theories to Adolescent Substance Use." The International Journal of the Addictions, 21, 3, pp. 347-366.

Winfree, L. Thomas, and Curt Taylor Griffiths (1983) "Social Learning and Adolescent Marijuana Use: A Trend Study of Deviant Behavior in a Rural Middle School." Rural Sociology, 48, 2, pp. 219-239.