Why do .students use marihuana? This question is really two different questions: Why do students begin to use marihuana and why do students become regular marihuana users? Two different answers are encountered; one psychological and the other sociological.

In 1958 David Ausubel, an authority in the field of drug use could state that a habitual marihuana user comes "from a poverty-stricken broken home or a home marked by domestic strife. Frequent nightmares, enuresis, and other symptoms of emotional tension and disturbed interpersonal relationships are typical of his early childhood history.' More recently a government booklet entitled Answers to the Most Frequently Asked Questions About Drug Abuse asked why people continue to use marihuana and concluded, "the consistent user, the `pothead,' is likely to be emotionally disturbed, according to many studies of this group. He is using the drug to treat his personality problems."2

This emphasis on the psychological aspects of drug use appears to be fundamental to the official position. Almost all statements from the medical associations, law enforcement officials, and drug "information" literature emphasize psychological factors.3 In spite of the official importance attached to psychological explanations, definitive empirical proof is lacking. A psychiatrist, Lester Grinspoon, examined many studies attempting to link personality factors with marihuana use. He found that most clinical studies in the United States or other countries are badly misleading or methodologically insufficient. He concludes that "there can be no doubt that certain personality types are more attached to the use of marihuana."4 But he does not specify what these personality types may be. Increasingly, surveys of college and high school students are beginning to demonstrate that various neurotic indices are associated with drug use.' Grinspoon points out that this kind of data must be interpreted with care. Many drug users are quick to admit confusion and anger in an attempt to be "open" to their shortcomings. They respond positively to items indicating psychopathology to which equally disturbed, but less open, students respond negatively.6 Another difficult problem is that many psychological scales are based upon conventional social values that drug users reject. Such personality measures are mainly a reflection of the dominant group ideology and yet are arbitrarily defined as indicators of mental health or pathology. Since many drug users disagree with the value assumptions expressed in questionnaire items, they provide answers that are interpreted by authorities as indicative of psychopathology. Given the problems with psychological scales, it is not surprising that marihuana use is associated with indicators of psychological difficulties. However, personality variables are probably not as strongly related to marihuana use as the social variables discussed below.

In contrast to the limited evidence supporting personality explanations of marihuana use, there are many large-scale methodologically sound surveys of drug use in nonclinical populations. Although such researchers seldom investigate personality factors, they manage to demonstrate that many social factors are strongly related to the use of marihuana and other drugs. Using survey research, Blum, Goldstein, and Smart present 25-100 different variables significantly related to marihuana use. Some of the variables these source's and others' have found to be related to marihuana are sex, socioeconomic status, religiosity, authoritarianism, alienation, peer and parental drug use, self-social identification, cigarette and alcohol use, familial instability, political activism, and premarital sexual involvement. To the layman and professional researcher, the implication of these statistical associations is that drug use is "a complex behavior with multiple causes."9

This "multiple cause theory" implies an additive model of drug use. If several of the variables correlated with cannabis use could be added together, one could predict with a high degree of probability whether a person will use marihuana or not. Cisin and Manheimer found that combining sex, marital status, cigarette use, religious preference, and the nonmedical use of prescription drugs allows one to predict marihuana use quite successfully. Among San Francisco area adults, 91% of those without children, with no religious preference, and having obtained a prescription drug from nonmedical sources had tried marihuana. At the other extreme, only 2% of the married females with children and having a regular doctor have tried marihuana." These findings are parallel to the findings reported below.

This study suggests that it really does not matter what background factors are used as the independent variables. Since the multiple cause theory holds that there are many other variables, as noted above, selecting and combining a few such factors increases the accuracy of predicting whether certain extreme social groups will or will not use cannabis. For this study, the four background factors that are most (not true for sex) highly correlated with marihuana use were selected. When these factors were applied simultaneously, almost 100% of those subject to all factors were shown to use cannabis. These variables are the respondent's religiosity, subjective political orientation, cigarette use, and sex.11

A brief discussion of three of these variables is necessary.

Our measure of religiosity is not intended to indicate differences in personal beliefs or theology. It measures the degree of affiliation with conventional religious institutions. Persons have been classified on the religiosity variable as follows. (1) "Not religious" if they are nonaffiliates or never had a religious preference. Most of these persons indicate that while they were raised as a Jew, a Catholic, or a Protestant, they presently have no Judeo-Christian preference. They have given up organized religion. (2) "Some religiosity" if the respondent maintains the religious preference in which he was raised but has greatly decreased his attendance at religious services since being a high school senior. Also included here are a few persons who have changed from one major faith to another. (3) "Very religious" if the respondent maintains the religious preferen& in which he was raised and attends religious services about as frequet,ttly as he did while a high school senior.

The variable "cigarette use" (which does not include smoking pipes or cigars) classifies respondents using cigarettes: (1) "daily"; (2) "less than daily" if the respondent had used cigarettes at least once after July 1, 1969, or smoked cigarettes more than monthly before July 1, 1969, but has not used them since that time; and (3) "never" if the person claims never to have used cigarettes or had used them less than monthly prior to July 1, 1969.

Respondents were classified on "political orientation" according to their own evaluation of where they stood on a continuum from "radical" to "very conservative." Persons were classified as follows: (1) "left" if they said they were "radical" or "very liberal"; (2) "liberal" if they said "moderately liberal," "indifferent," or "don't know" or if they refused to answer or left the question blank;" and (3) "moderate" if they considered themselves "moderate," "moderately conservative," or "very conservative."

The dependent variable is the Frequency of Cannabis Use Index, described in Chapter 3. For the most part, experimental (less than monthly) and moderate (less than weekly) users are not considered because the concern here is to analyze why students ever try and become regular (weekly) users of cannabis.

The relationship between these four background variables and the use and regular use of cannabis is found in Graph 4.1. At the top of each section the strength of the relationship is noted. Hence, the percent difference in the use of cannabis (symbolized by %U) between males and females is 15% (63%-48%). In addition, the sex difference in regular cannabis use (%R) is 12% (25%43%). Other factors are even more strongly related to cannabis use than sex. Nonreligious persons are 43% (%U = 75%-32%) more likely to ever use cannabis and 24% (%R = 31%-7%) more likely to be regular cannabis users than very religious persons. The 46% difference between daily smokers versus noncigarette smokers in marihuana use is virtually identical to the 47% (50%-3%) difference, reported by Josephson, in the national high school population.13 The 55% difference between those who consider themselves left, versus moderate, is somewhat stronger than the 40% difference reported by Gallup in a national college population." But the direction of the relationship is the same in this survey as in other major surveys. The most important factor, discussed in much greater detail below, is peer-group cannabis use. Respondents who claim that most of their friends use cannabis are 74% (90%-16%) more likely to use marihuana themselves than persons who claim to have few cannabis-using friends.

The basic intention here is to demonstrate the validity of the multiple cause theory by showing that four background variables (sex, religiosity, cigarette use, and political orientation) predict with great precision whether students will use marihuana or not. An attempt will be made to show that each background variable is additively and independently related to the use of cannabis. In Graph 4.2, these'variables are held constant; different combinations of religiosity, smoking, and political orientation for men (p. 56) and women (p. 57) are presented.

For those students in the extreme groups one can predict with virtual certainty the use or nonuse of cannabis, because of the independent and additive effect of these four background variables. Among the 119 nonreligious, politically left, daily cigarette-smoking males (upper left), 97% have tried cannabis and 62% are weekly cannabis users. At the extreme lower right, only 4% of the 180 very religious, politically moderate, noncigarette-smoking females have ever tried cannabis and none are weekly cannabis users. These findings are not the result of statistical accidents; the numbers of cases (119 and 180) are sufficient to provide stable percentages. This finding is parallel to the Cisin and Manheimer data cited previously.

In addition, these data appear to be the result of social forces at work. Each of the four factors contributes independently and in relatively predictable ways to the use and weekly use of cannabis. Although there are a few exceptions, the level of marihuana use and the regular use of cannabis declines steadily as one goes from the upper left corner to the lower right corner of Graph 4.2.

As expected from the multiple cause theory, by combining only four background factors one can predict with great precision whether a person will or will not use cannabis. The important thing about these four variables is that they are essentially social variables. One need not consider personality or psychological factors to be extremely accurate in predicting marihuana use. There is no need to hypothesize a drug-prone personality or psychological disturbance. This does not mean that personality factors may not be important, but they are not necessary to understand marihuana use.

Nevertheless, the above multiple cause theory is greatly misleading. Howard Becker, as long ago as 1953, recognized the fallacy of this theory. What we call the multiple cause theory, Becker called predispositional theory. He felt that a theory built upon such background variables was basically inadequate because such "predispositional theories cannot account for that group of users (whose existence is admitted) who do not exhibit the trait or traits considered to cause the behavior and such thebries cannot account for the great variability over time of a given individual's behavior with reference to the drug.15

Instead, Becker's early writings suggest that involvement in marihuana-using circles was essential. Becker indicates that a marihuana-naive person must learn from someone, usually a friend, how to smoke the marihuana cigarette and how to recognize and enjoy the effects of the drug. In order for the person to become a regular marihuana user, the person must insulate himself against pressures not to use marihuana, which come from the definitions of morality of nonusing friends, family, and public. Insulation is usually facilitated by disaffiliation with family and friends who frown upon use and reaffiliation with persons who use and rationalize the consumption of cannabis and other drugs. 16

Although the concept of insulation is used or suggested by Becker and others," no writer carefully defines what is meant. This study will -define "insulation" as occuring when almost all of one's friends engage in a particular activity—in this chapter, when almost all friends use marihuana. Evidence indicates that a person is not insulated if even 25% of his friends are nonusers. Having a few nonusing friends is probably the most important factor limiting the regular use of cannabis, although some marihuana use may occur.

In order to test Becker's ideas, respondents were asked, "Among your present friends, about how many have tried marihuana?" Respondents were classified as having: (1) "few" if less than a quarter of their friends used marihuana; (2) "some" if a quarter but less than three-fourths of their friends used it; and (3) "most" if almost all or all of their friends used cannabis. This variable, peer-group cannabis use, appears to be a good indicator of insulation from nonusing friends and involvement in marihuana-using circles.

This variable is really a measure of perception by the respondent of his friends' use of marihuana. It is not a measure of whether his friends actually use marihuana or not. However, with a highly visible activity such as marihuana use, there is likely to be a high correlation between the respondent's perception of, and his friends' actual use of, cannabis. Hence, peer-group cannabis use variable appears to be a good indicator of involvement in cannabis-using groups. Graph 4.1 demonstrated that those with most of their friends using cannabis were 74% more likely to try cannabis than those with few cannabis-using friends. In addition, weekly cannabis use is common (41%) among those with most of their friends using marihuana. Respondents with some cannabis-using friends (9%) are much less likely, as are those with few friends (1%), to become regular cannabis users. The data seem to demonstrate, therefore, that having some cannabis-using friends is conducive to trying cannabis but not to regular use. For weekly cannabis use to occur, the user must insulate himself in a group where most friends use cannabis. Exposure to even a few (about 25%) friends who do not use cannabis appears to inhibit strongly regular cannabis use.

Having briefly described the variable peer-group cannabis use, the study will now attempt to demonstrate the weakness of the multiple cause theory.

Although not presented, data show that each of the four background variables is significantly and strongly related to peer group cannabis use when used as a dependent variable. For example, 55% of the nonreligious versus 17% of the very religious respondents have friends most of whom use cannabis. Similar findings hold for political orientation: 61% of left versus 14% of moderates have most friends using cannabis. Hence, these four background factors greatly increase the probability that a student will have cannabis-using friends. If Becker's logic is generally correct, the relationship between each of the background factors and marihuana use should be reduced when peer-group cannabis use is held constant. This is precisely what happens. For example, Graph 4.3* shows that the religiosity difference (%U = 43%) in cannabis use is reduced to an average percent difference (APD) of 17% when peer-group cannabis use is held constant. This tends to indicate that the strong relationship between religiosity and cannabis use exists mainly because less-religious persons become involved in peer groups where cannabis is used and hence use the drug themselves. The direct effect of religiosity on cannabis use is much less important. Similar findings hold for the other factors. Thus, the 46% difference in cannabis use due to cigarette use is reduced to 25% and the 55% difference due to political orientation is reduced to 25% when peer-group cannabis use is held constant. Only the sex differences are not greatly reduced. Thus, the 15% difference in cannabis use between males and females is reduced to 7% and 3%, respectively, among those with friends few and most of whom use cannabis. However, among those with some cannabis-using friends males are 16% more likely to use cannabis than females. This may indicate that when persons are pressured by some friends who
want them to use cannabis and by other friends who would prefer that they avoid it, males will try cannabis and females will avoid it.

Graph 4.3 also allows a better understanding of regular cannabis use. For example, the 27% difference in regular cannabis use due to daily versus noncigarette use appears to be reduced by half (APD = 14%) when peer-group cannabis use is held constant. However, this is misleading. Among those with most of their friends using cannabis, daily cigarette users are 27% (%r = 52%-251) more likely to be regular cannabis users than noncigarette users. This tends to demonstrate that in cannabis-using groups, being a noncigarette user is associated with inhibiting the regular consumption of cannabis. Similar findings emerge from the analysis of other background variables. Background characteristics that mitigate against cannabis use (female, political moderation, and very religious) tend to inhibit the weekly use of cannabis, especially among those who have most of their friends using cannabis. Among those with few or some cannabis-using friends, the background characteristics seem to have much less influence in curbing or increasing the probability of regular use.

The basic weakness of the multiple cause theory can now be dealt with. The four background variables are accurate in predicting cannabis use, because they are closely associated with having cannabis-using friends. Graph 4.2 showed that there were 119 persons who were nonreligious, daily cigarette-using, politically left males. Two of these persons did not indicate their friendships with cannabis users. Of the remaining 117, 76% claimed that most of their friends use cannabis; 18%, some friends; and 6%, few friends. At the other extreme, one respondent did not indicate how many friends used cannabis. Of the remaining 179 very religious, noncigarette-using, politically moderate females, 86% have few, 12% have some, and only 2% have most friends using cannabis. Hence, these background variables are important in determining whether persons will have cannabis-using friends or not.

Once something is known about the cannabis use of a person's friends, knowing a great deal about the background variables does not greatly increase the ability to predict cannabis use. Thus, the primary factor in understanding why a college student used marihuana is the proportion of his friends who use it. The important sociological question becomes, Why do students have marihuana-using friends? An adequate answer to this question is beyond the scope of this book. However, the multiple cause theory suggests that background variables, especially when several are combined, increases greatly the probability of having many cannabis-using friends.

Nevertheless, the background variables have a slight effect on cannabis use that is not explained by friend's cannabis use. Graph 4.4 presents information about those 117 persons who are most likely and the 179 persons who are least likely to use cannabis while holding constant peer-group cannabis use. The most impressive feature of the graph is the amazing accuracy with which cannabis use is determined by the five independent variables. Thus, 100% of the 89 nonreligious, daily smoking, politically left males with most friends using cannabis have themselves tried cannabis and 74% are regular cannabis users. At the other extreme, only 2% of the 154 very religious, nonsmoking, politically moderate females have tried cannabis and none are regular cannabis users. A difference of this magnitude, 100% versus 2%, is seldom seen in the social science literature. Nor, to the author's knowledge, has other research been able to predict as accurately why people are regularly involved in a particular behavior or not (74% versus 0%). In short, cannabis use and regular use is a highly predictable phenomenon. Given information on just five predominatly sociological variables, one can predict with a high degree of accuracy whether a person will or will not use cannabis.

Graph 4.4 also shows that the background factors have some effect independent of friends. Although there are not enough cases for stable percentages, the results are suggestive. Among students who are likely to try cannabis (nonreligious, daily smoking, left males) but who have few or some cannabis-using friends, most have tried cannabis (71% and 95%, respectively) but are unlikely to be regular cannabis users (14% and 23%). Perhaps such persons participated in cannabis-using circles at an earlier time but they were not involved in such groups when completing the questionnaire. If this is the case, there is a need to analyze such change over time.

At the other extreme, those least likely to use cannabis (religious, nonsmoking, moderate females) are unlikely (19% and 25%) to use cannabis even though some or most friends use cannabis; however, the numbers of cases (21 and 4) are too small for generalization. Having a very conventional life style appears to inhibit the use of cannabis, even among those who have several cannabis-using friends.

For the most part, the discussion has been confined to extreme groups and has ignored the 90% or more of the population that is not so highly predictable. There is an important reason why cannabis use is not so easily determined for a majority of the students. While many persons are exposed to social pressures that should lead to drug use, they are exposed at the same time to pressures to avoid drug use. For example, evidence not presented here shows that daily cigarette smoking, politically liberal, very religious females have intermediate rates of cannabis use (66%) and mixed friendships (35% have few, 28% have most friends using cannabis). Thus, many persons are exposed to conflicting pressures that hold down the level of marihuana use. But, as noted in Graph 4.2, the general trend is clear: when various combinations of background factors are known, both the level of marihuana use and friendships with cannabis users are generally predictable within certain limits.

The data indicate that background factors are linked to cannabis use'by the cannabis use of one's friends, but it has not been established which factor, peer-group cannabis use or the respondent's use of cannabis, is predominantly causal. According to Becker, a person is generally introduced to cannabis by a friend. However, in order for marihuana use to become regular, the prospective user must increasingly insulate himself from those who would disapprove. This insulation, if it occurs, is necessary before heavy marihuana use is possible. Furthermore, increasingly regular cannabis use is facilitated the more one becomes friends with cannabis users.18

Fortunately, there is a technique available by which one can determine the time order of cannabis use and having friends who use marihuana. The basic methodological technique is the sixteenfold table." In the questionnaire respondents were asked to report their marihuana use and friends' use during their senior year in high school. There is also data, of course, on the use of cannabis and friends' use at the time of questioning. Four variables were developed and dichotomized in the following manner: (1) marihuana use in high school, yes or no; (2) present use of cannabis, yes, if the students had used it after July 1, 1969, and no, if only before July 1, 1969, but not since or never; (3) peer-group cannabis use in high school, "few" if less than a quarter of friends used it and "many" if more than a quarter of friends used marihuana; and (4) peer-group cannabis use at present, the same code as for high school use.

The basic logic of a sixteenfold table is that it allows the researcher to examine the relationship between the same two variables at two different points in time. By doing this the researcher can determine which of the two factors is most important in determining the other factor. Table 5 permits a test of whether friendships with marihuana users are shifted to accord with the respondent's use or whether his marihuana use changes because of peer-group pressures. In this kind of analysis raw numbers are used rather than percentages.

In Table 5, one cannot determine which variable is the cause of the other by examining the cells on either diagonal, since the upper left to lower right diagonal (1117, 18, 150, 393) is composed of persons who have not changed, while the lower left to upper right diagonal (2, 2, 34, 706) is composed of those who have changed on both variables between high school and college. It should be noted that the direction of changes is to gain friends and begin cannabis use (706 and 34); very few (2 and 2) give up friends or cannabis. By examining the other eight cells in the table, one can determine which of the variables have the greater "generating" and "preserving" effects.

Let us first examine the generating effect. This effect occurs when persons who are inconsistent on the two variables at time 1 (in high school) become consistent at time 2 (the present). For example, persons who are in the middle two rows of Table 5 are inconsistent in their friendships or cannabis use in high school. Those who became consistent at time 2 are found in the first and fourth columns.

If one assumes that pressure from friends determines the respondent's use, then high school cannabis users with few marihuana-using friends should give up their cannabis use; there are 17 such persons. Furthermore, nonusers in high school with many marihuana-using friends should take up marihuana later; there are 119 such persons. Thus, 136 (17 + 119) persons have adjusted their marihuana use to their friends' cannabis use. For these persons, pressure from friends "caused" the respondent's nonuse or use of marihuana.

Now assume that the respondent's marihuana use determines his friendships with cannabis users. If this is the case, nonusers in high school with many marihuana-using friends should give up such friends; there are 33 such persons. Further, high school marihuana users having few cannabis-using friends should acquire many cannabis-using friends in college; there are 197 such persons. Thus, for 230 (33 + 197) persons, their own marihuana use has "caused" them to drop or gain cannabis-using friends. Hence, the predominate direction of causality (230 versus 136) is the latter outcome. Persons change their friendships to accord with their own drug use rather than alter their use of drugs because of pressure from friends. The preserving effect holds that persons who are consistent at time 1 will tend to give up the least powerful factor by time 2, thus "preserving" the strongest factor. Essentially, Table 5 shows that more persons gain or give up their friendships to accord with their own marihuana use (8 + 537) than shift their marihuana use to accord with their friendships (85 + 78). Thus, for both the generating and preserving effects the respondent's own use of cannabis is a more important determinant of having cannabis-using friends than vice versa. Although the predominant tendency is for friendships to be adjusted to the respondent's own use, one must not forget that in many cases the opposite occurs. This tends to support Goode's assertation that

a dynamic and dialectical relationship obtains between friendships and the amount one smokes. One does not generally acquire a great many marijuana-smoking friends until one already smokes. Yet the fact that one has friends who smoke further increases the likelihood that one will smoke more. . . . Rather than one or another being causal here, the two mutually influence and feed into each other."

The evidence presented in this chapter indicates that cannabis use is a highly predictable phenomenon when four background variables (sex, religiosity, political orientation, and cigarette use) and peer-group cannabis use are combined (five independent variables); extreme groups are almost perfectly predicted. Thus, of the 89 (2.5% of the total sample) nonreligious, left, daily smoking males with most of their friends using cannabis, 100% have ever used marihuana and 74% are weekly cannabis users. At the other extreme, of the 154 (4.5% of the sample) very religious, moderate, nonsmoking females with few cannabis-using friends, only 2% have ever used marihuana and none are weekly cannabis users.

In addition the data help integrate the multiple cause theory with Becker's theory of social interaction in explaining cannabis use. Background factors strongly influence the probability of participating in peer groups where cannabis is utilized. However, membership in a peer group where marihuana is used, no matter what the background factor, greatly increases the chances that the individual will begin to use marihuana.

Once students begin using marihuana, two major factors affect the frequency of use. First, the proportion of the peer group using marihuana will greatly affect the frequency of use. If even 25% of one's friends do not use marihuana, the respondent's use of cannabis will tend to be irregular. However, if almost all or all friends use it, the respondent's use will become increasingly regular. Furthermore, there appears to be a tendency for novice marihuana users to seek new friends and social circles where almost everyone uses marihuana.

Second, the background factors also affect the regular use of cannabis. Each background factor increases the likelihood of gaining cannabis-using friends and hence progression to regular cannabis use. If the background variables indicate that persons should not use marihuana, respondents will be unlikely to have cannabis-using friends and hence unlikely to become a regular user. In addition, students having most friends using cannabis are unlikely to become regular marihuana users if one or more of the background variables indicate that students should not use cannabis.

These findings have important implications for drug subculture theory. First, the data demonstrates the enormous importance of social factors in determining marihuana use. This study has not resorted to psychological explanation to understand cannabis use. While the importance of personality factors has not been dispmven, their importance is challenged. The present logic suggests that persons who are emotionally disturbed (however this is defined) are unlikely to try marihuana if few of their friends use it, while psychiatrically normal persons with cannabis-using friends are likely to try it.

Second, the basic conclusion of this chapter is that the most important factor in explaining the respondent's marihuana use is the use of cannabis among his friends. Among those with most friends using cannabis, 90% have used it; the 10% who have not tried marihuana are restrained by other social factors, such as noncigarette use or a moderate or liberal political orientation. Hence, the peer-group cannabis use variable is an important measure of what is meant by participation in the drug subculture. This variable measures the effects of social interaction centered around the consumption of cannabis. The greater the involvement of students in cannabis-using circles, the more students are subject to the conduct norm, "Thou shalt use marihuana." Very few students can resist this insistent, but supportive, pressure from friends to use marihuana unless they are exposed to other pressures against marihuana use from their background and general environment; even so, friends usually win.

Third, if one views the four background variables as indicators of commitment to the parent culture, one can understand how this culture is connected to marihuana use and, by inference, to initial and increasingly regular participation in the drug subculture. The parent culture structures and defines appropriate behavior for those occupying various statuses in society. Such statuses, or background factors, become important selector criteria for membership in cannabis-using peer groups and the drug subculture. In a similar fashion, background factors or statuses, such as the parental socioeconomic status, alcohol use, Peer Culture Index, familial instability, alienation, and parental drug use, which have not been dealt with here, are connected to the drug subculture.

If cannabis use and initial drug-subculture participation is the result of social interaction that is structured by the behaviors and ordinary social statuses of college students, it is difficult to imagine how laws and governmental action may greatly alter such patterns of use. Few government or college officials in the United States would try, for example, to get students to go to church. Nor could laws greatly affect political liberality. Indeed, many colleges in this study are proud of their role in instilling liberal values in their students. While many persons would like to end cigarette smoking, the policy on campus and elsewhere is to tolerate smoking; laws against smoking by minors are almost completely unenforced. And finally, laws can do little to alter the patterns of friendship formation, especially in the direction of nondrug use.

Although drug laws are intended to prevent cannabis use, the evidence here indicates that the opposite effect is more probable. Present drug laws, by defining drug use as a criminal act and promoting the idea that cannabis use is immoral, may drive marihuana use underground, so that users give up nonusing friends and insulate themselves in circles of cannabis-using friends. In addition, such laws may regularize cannabis use by insulating cannabis users from nonusers who might reduce the respondent's use of cannabis.

While it is hard to demonstrate convincingly and empirically that the drug laws are important in determining the use and regular use of cannabis, it is far easier to demonstrate that the illegality of cannabis plays a crucial role in the illegal sale of marihuana and hard drugs, the topic of the next chapter.

*This Highly complex graph demonstrates the kind of analysis that will be used throughout the rest of the book. Readers are urged to study carefully this footnote to understand how the relationship between an independent variable and a dependent variable is altered by holding constant a test factor. One must first consider the strength of the relationship between the independent and dependent variables (sociologists refer to this as the original, or two-way, relationship). A simple and effective way to measure the strength of a two-way relationship is to calculate the percent difference. The percent sign [AU] before a letter symbolizes the percent difference between extreme catagories of the independent variable; the capitalized letter symbolizes the dependent variable (e.g., U = cannabis use) and the two-way relationship between an independent variable and this dependent variable. Thus, the %R is the percent difference in regular cannabis use due to the independent variable. For example, using religiosity as the independent variable, Graph 4.1 shows that nonreligious persons are 43% (%U = 75%-32%) more likely than very religious persons to have ever used cannabis and 27% (%R = 31%-7%) more likely to be regular cannabis users.

Why the nonreligious are more likely than the religious to use and to use regularly cannabis can be shown by holding constant the test variable peer-group cannabis use. This variable is held constant by splitting the sample into three subgroups, those with few, some, and most cannabis-using friends. Within each subgroup of the variable, the strength of the relationship between religiosity and drug use is examined. Lower-case letters (%u and %r) stand for the strength of the relationship between independent and dependent variables within a subgroup. Thus, among those most of whose friends use cannabis, the nonreligious are slightly (%u = 94%-80% = 14%) more likely than the very religious to try cannabis. Religiosity has some effect upon cannabis use among those with some cannabis-using friends 14% + 17% + 21%

This 17% has a specific meaning, which is easily related to the original relationship. The strong L%U = 43%1 relationship between religiosity and cannabis use is reduced to 17% when peer-group cannabis use is held constant. Another way of understanding the meaning of the APD is that religiosity has an independent effect of 17% upon cannabis use when peer-group cannabis use is held constant. For all practical purposes, religiosity has an independent effect which is about a third (17%/43%) of its original strength. What this means is that having cannabis-using friends, and not religiosity, is the important factor in determining cannabis use. Generally a reduction of two-thirds to 50% from the original relationship is sufficient to arrive at this conclusion.

In the following chapters and graphs the percent differences among subgroups [%u's and Yves] will be eliminated because they make the graphs confusing and are not central to the argument. For the most part, the original relationship (%U) will be presented and compared to the APD holding constant some theoretically important independent variable.

REFERENCES

1. David P. Ausubel, Drug Addiction, New York: Random House, 1958, p. 99.
2. Answers to the Most Frequently Asked Questions About Drug Abuse (pamphlet), Washington, D.C.: Government Printing Office, 1970.
3. Nathan B. Eddy et al., "Drug Dependence: It's Significance and Characteristics," Bulletin of the World Health Organization, 32 (1965), 721-733. American Medical Association, "Dependence on Cannabis," Journal of the American Medical Association, 201 (Aug. 7, 1967), 368-371; "Marihuana and Society," ,AMA, 204 (June 24, 1968), 91-92. Jerome LaBarre, "Harms Resulting from the Use of Marihuana," The Prosecuter, 6 (Mar.-Apr. 1970), 91.
4. Lester Grinspoon, Marihuana Reconsidered, Cambridge, Mass.: Harvard University Press, 1971, pp. 253-290.
5. Ronald A. Steffenhagan and Patrick J. Leahy, "A Study of Drug Use Patterns of High School Students in the State of Vermont," unpublished, University of Vermont, 1968. C. P. McAree et al., "Personality Factors in College Drug Use," International Journal of Social Psychiatry, 15(2) (1969), 92-106. Joel Goldstein et al., The Social Psychology and Epidemiology of Student Drug Usage: Report on Phase One, Department of Psychology, Carnegie-Mellon University, June 1970, pp. 67-71. Ulla Ahrens, Magnus Kihlbom, and Nils Nas, "Students and Narcotics: An Investigation of Students at the University of Stockholm in Spring, 1968" (translated title), in Statens offentliga utredningar, Socialdepartementet, Narkotikaproblemet ("The Drug Problem, Part IV, Sociomedical and Clinical Investigations"), Stockholm, Sweden, 1969, pp. 173-178. Richard Blum, Students and Drugs, San Francisco: Jossey-Bass, 1969, pp. 233-241. Robert Hogan et al., "Personality Correlates of Undergraduate Marijuana Users," Journal of Counseling and Clinical Psychology, 32(1) (Aug. 1970), 58-63.
6. Grinspoon, Ref. 4, p. 286.
7. Erich Goode, Drug Escalation: Marijuana Use as Related to the Use of Dangerous Drugs, Paper prepared for the National Commission on Marihuana and Drug Abuse, Oct. 30, 1971, p. 50.
8. Blum, Ref. 5, pp. 63-167. Goldstein et al., Ref. 5, pp. 18-48; Reginald Smart et al., "Preliminary Report on the Attitudes and Behavior of Toronto Students in Relation to Drugs," Toronto, Canada: Addiction Research Foundation, Jan. 1969. Bruce Johnson, Social Determinants of the Use of "Dangerous Drugs" by College Students, Ph.D. thesis, Columbia University, privately published by author, June 1971, Appendix B. Ahrens, Kihlbom, and Nas, Ref. 5, pp. 109-192. Gallup International, "Special Report on the Attitudes of College Students," Gallup Opinion Index, Report No. 48, Princeton, N.J., 1969. Leonard Goldberg, "Drug Abuse in Sweden," Bulletin on Narcotics, 20 (1968), 1.31. Kenneth Eells, "Marijuana and LSD: A Survey of One Campus," Journal of Counseling Psychology, 15(5) (1968), 459-467. Lillian Imperi et al., "The Use of Hallucinogenic Drugs on Campus," Journal of the American Medical Association, 204 (June 17, 1968), 1021-1024. Paul M. Kohn and G. W. Mercer, "Drug Use, Drug Use Attitudes, and the Authoritarianism-Rebellion Dimension," Journal of Health and Social Behavior, 12 (June 1971), 125-131. Lillian Robbins et al., "College Students Perception of Their Parents; Attitudes and Practices Toward Drug Use" (unpublished), New York, 1970. Edward Suchman, "The Hang-Loose Ethic and the Spirit of Drug Use," Journal of Health and Social Behavior, 9 (June 1968), 146-155. Lawrence S. Linn, "Social Identification and the Use of Marijuana," International Journal of Addictions, 6 (Mar. 1971), 79-107.
9. Joel Goldstein, "Motivations for Psychoactive Drug Use Among Students," Paper presented to Eastern Psychological Association, New York, N.Y., April 15, 1971.
10. Ira H. Cisin and Dean I. Manheimer, "Marijuana Use Among Adults in a Large City and Suburb," Paper presented at the New York Academy of Sciences, May 21, 1971. An earlier report of the San Francisco sample indicates a fairly high (.62) multiple correlation coefficient and that religious affiliation is one of the single most important variables: Dean I. Manheimer et al., "Marijuana Use Among Urban Adults," Science, 166 (Dec. 19, 1969), 1554-1555.
11. In personal conversations with the author, Eleanor Carroll of NIMH expressed a desire to know more about the drug use patterns of females. Hence, we have included sex as an independent variable, although it is not as highly associated with marihuana use as the other three variables. On the basis of our evidence, we feel that females behave almost the same but have slightly lower levels of marihuana use than males.
12. The question on political orientation was at the end of the questionnaire; about 425 students left it blank, were indifferent, or could not classify themselves. Since we wish to compare extremes (left versus moderates) and include all cases for detailed analysis, these 425 cases have been included in the intermediate "liberal" category. It must be noted that our New York City sample is more "left" than the national college population. Comparison with Groves et al., "Study of Life Styles and Campus Communities: Preliminary Report," Johns Hopkins University, Dec. 1970, shows that 9% of our sample, but 5% of the national college population, consider themselves radical. About 27% of our sample, but 56% of the national college students, consider themselves moderate or conservative.
13. Eric Josephson et al., "Adolescent Marijuana Use: Report on a National Survey," Paper presented to the First International Conference on Student Drug Surveys, Newark, N.J., Sept. 14, 1971, p. 9.
14. Gallup, Ref. 8, p. 30.
15. Howard S. Becker, "Becoming a Marihuana Smoker," American Journal of Sociology, 59 (Nov. 1953), 236.
16. Ibid., pp. 235-242. Howard S. Becker, "Marihuana Use and Social Control," Social Problems, 3 (July 1955), 35-44.
17. Erich Goode, "Multiple Drug Use Among Marijuana Smokers," Social Problems, Vol. 17, Summer 1969, pp. 54-62. James Carey, The College Drug Scene, Englewood Cliffs, N.J.: Prentice-Hall, 1968. Alan G. Sutter, "Worlds of Drug Use on the Street Scene," in Donald Cressey and David A. Ward, eds., Delinquency, Crime and Social Processes, New York: Harper and Row, 1969, pp. 802-829.
18. Becker, Ref. 15, p. 237; and Ref. 16, p. 35. Goode, Ref. 17, p. 56.
19. Paul F. Lazarsfeld, "Mutual Effects of Statistical Variables" in Paul F. Lazarsfeld, Ann K. Pasanella, and Morris Rosenberg, Continuities in the Language of Social Research, New York: Free Press, 1972, pp. 388-398. This kind of analysis is based upon having data from the same respondents at two different points in time (panel data). The data we report here is pseudo panel data because we use retrospective data, asking the respondent to report his activity in the past. Thus, our findings may be misleading.
20. Goode, Ref. 17, p. 56.