This chapter reports an exploratory effort to predict which students will begin illicit-drug use during the academic year. It is based upon the p-remise that among the many variables related to initiation of drug use, those which will operate as influences are past drug-taking experiences, expressed willingness to take drugs, the past illicit-exotic drug-taking experiences of one's chosen friends and admired peers (leaders), and the willingness to take drugs expressed by friends and leaders.

Subjects were asked to rate their willingness to take drugs under two conditions: a medical-research experiment to evaluate the effects of the drug, and a social situation where the drug is available without cost from a friend or an acquaintance. Willingness was to be rated on a scale of five from definitely unwilling through neutral to definitely willing. Eighteen drugs were listed. These included as follows: three hallucinogens, psilocybin, LSD, and peyote; two sedative-hypnotics, phenobarbital and seconal; three narcotic painkillers, Demerol, heroin, and codeine; one tranquilizer, Miltown (Equanil) ; two cannabis preparations, marijuana and hashish; two accepted social drugs, alcohol and tobacco; three stimulants, benzedrine, dexedrine, and methedrine; and two fictitious drugs, ergathol and TM-15. The list indicated the classification of each drug except marijuana, tobacco, and alcohol. The fictitious drugs were described as having "varied effects." To develop a score, the sum of willingness scores for all drugs under each condition was summed and averaged. When there was an interest in willingness by type of drug, for example, hallucinogens, the average willingness score for drugs in this class was obtained. In cases where a subject failed to indicate willingness, the willingness scored was based on the average of ratings given.

For each of the eighteen drugs, subjects were asked to indicate the approximate number of times they had taken the drug, including medical and nonmedical uses. When exact figures were unlikely to be known, as, for instance, in the case of alcohol or tobacco, frequency in terms of guesses or estimates was requested.

An attempt was made to validate the willingness scale, although it might just as well be considered a measure of consistency or reliability. An accomplice was selected from among male students1 living together in a dormitory, all of them enrolled in the same summer language program. The accomplice named seven fellow residents with whom he could speak frankly about drugs. Each of these seven plus the accomplice was independently approached by the experimenter and told he had been selected at random to take the willingness scale. In the week following, the accomplice approached each of the other students and initiated a drug discussion in which he individually asked them about their willingness to take marijuana, benzedrine, and phenobarbital at a get-together where the drugs would be available. In the hypothetical situation posed, each subject was asked whether he would take such drugs if someone present proposed doing so to get "high." After these discussions, or "invitations," the accomplice rated each subject on the willingness scale. The twenty-one ratings (seven subjects, three ratings each) were compared with the corresponding self-expressions of willingness originally obtained. The association between the friend's rating and the subject's self-rating yielded r = .78, t = 29.5 (P < .01), which shows a positive correlation between stated willingness expressed in a formal setting ("tested" by an unknown student researcher) and willingness rated by a friend after a discussion of drug taking in a setting which implied an invitation to take drugs.

Several expectations about the relationship of stated willingness to other variables were set forth and tested on a student sample. The sample was casually drawn from students going through registration at a junior college. The experimenter, himself long-haired and casually dressed (in hippie garb, according to some), asked each of seventy-five students (consecutive) to participate. Thirteen refused, eight failed to complete the questionnaire, and fifty-four remained, of whom thirty were male and twenty-four female.

The range of possible scores on the Social Willingness Scale is 18 to 90. Actual scores obtained ranged from 18 to 86 with a median of 31. The first expectation tested was that social willingness would increase the more the related experience; thus, there would be least social willingness to take drugs not previously taken (Wn), more willingness to take those previously administered by a physician (Wd), and most willingness to take ones already previously used socially (Ws) —that is, procured through a friend or acquaintance. The scores, obtained by analyzing and then matching each subject's willingness and one-time-or-more experience with scale items, support the expectation. The'average social willingness for Wn drugs is 1.82, for Wd drugs is 2.45, and for Ws drugs is 3.07. Note that this latter score is only a bit above neutral, which indicates that social willingness is not predictable solely from past social use of the same drugs.

Illustrative of the factors contributing to differential willingness are differences by drugs and by amount of experience. For example, among those who have never taken marijuana (N = 32/54), Ws is 1.2; for those who have smoked it once (N = 6), Ws is 3.2; whereas for those who have smoked it more than once, Ws is 4.1. For benzedrine, on the other hand, those who have not used it average Ws of 1.4 and both one-time and more-than-one-time users average 3.3. LSD willingness (Ws) for inexperienced subjects (N = 46) is 1.7, whereas for those six subjects who have had LSD—and all have taken it at least three times—Ws is 4.5. For alcohol those who have taken it rarely (ten times or fewer) average Ws of 2.0, whereas more experienced drinkers express an average willingness of 3.9. Note that the range of scores is greatest for the two illicit-exotic drugs, ranging from relatively strong unwillingness among inexperienced persons to strong willingness among more experienced ones, whereas with the more acceptable drugs (even though alcohol is illegal in this under-twenty-one age group and benzedrine can only be legally obtained by prescription) lesser ranges of average willingness are obtained. One wonders whether, under conditions of broader sampling, the moderate-vs.-the-extreme range of social-willingness scores might not parallel the moral-attitudinal intensity surrounding particular drugs.

The range of possible scores on experimental willingness is also 18 to 90 and actual scores range from 18 to 87 with a median of 38 —a median higher than that obtained for social willingness. We would expect that social willingness and experimental willingness would be related and when tested this is found to be the case, r = .73 significant beyond P < .01. Controlling for past experience by excluding all drugs previously taken, one finds that a relationship still obtains, r = .60 (significant at P < .01), which means that there is a relationship between one's willingness to take a new drug socially and one's wlllingness to take it experimentally. The occurrence of that relationship should not let us overlook the large amount of variance for which we cannot account.

The questionnaire contained a number of other items which we thought would be associated with varying degrees of expressed willingness to take drugs. These covered the number of friends who have used tobacco, alcohol, marijuana, LSD, or other drugs for kicks, family use of medicines and drugs, involvement in typical social commitments, rejection of parents' values, rejection of society's values, general interest and curiosity, and orientation to life. We find significant correlations to obtain between the number of friends using the drugs listed and social willingness, r = .53, and between expressed interest and curiosity' and social willingness, r = .42. No significant relations are found between social willingness and the other factors as measured here.

A special inquiry was made into orientation to life, defined in terms of the ranked importance of the following themes: success, religion, politics, self-understanding, health, tradition, adjustment, having a good time, and family and children. (These same themes were the foci of ratings of parent-subject agreement on values and of society-subject agreement or disagreement on values.) Subjects with low, intermediate, and high social willingness were compared as to the frequency with which each theme was rated as "very," "slight," or "not important." A theme score for each Ws group was derived by subtracting the "slight" and "not-important" ratings from the number of "very-important" ratings. Among subjects with low social-willingness scores, health and adjustment rank higher than among the intermediate Ws group, which, in turn, ranks these higher than the high Ws group. Conversely, "having a good time" is ranked low by the low social-willingness group, whereas both intermediate and high Ws subjects rank it high. It is of interest that the themes of religion and tradition are uniformly ranked low by all groups.

From the foregoing data it appeared that the factors—among those examined—most likely to predict who would, in fact, "turn on" to illicit-drug use were expressed social and experimental willingness, past drug experience, illicit-exotic drug use among one's friends and, secondly, admired peers, and expressed willingness to take drugs socially. on the part of those same friends and group leaders.' We did not expect that these would be the only variables related to turning on, far from it, since drug access, personality variables, unusual social pressures, accident, and a host of other factors are likely to play a role. On the other hand, we expected that, given a situation in which drug use on campus was increasing, experience, willingness, and friendship patterns ought to play a detectable role.

The original plan called for enlisting the support of all members in two residential ( Greek-letter) fraternities, one high in illicit-exotic use, according to campus opinion, and one low. At the time the study began, the highest-use house selected was bathed in the unflattering light of various investigations of drug use by deans and other administrators, so that its officer, sensitive to further explosions, declined to cooperate. Two presumably low-use houses were willing to cooperate. The plan called for the experimenter to explain the study and to distribute and collect questionnaires at the beginning of the fall quarter and again at the end of spring quarter. Resistance in both houses to an outsider's playing a role meant inaugurating a new procedure, one in which a student resident in each house was enlisted to "manage" the work in each house. No direct supervision of him was possible since there was insistence on absolute anonymity of respondents and of freedom for nonresponse. A further sampling problem arose from the fact that in both houses some students were at overseas campuses in the fall while others were overseas in the spring. A further limitation arose when students completing both questionnaires selected as friends or leaders men who did not complete the initial questionnaire. Since the predictions rested upon knowing the drug experiences and willingness to take drugs of all those designated by a subject as friends or leaders, any missing questionnaire might be a needed one. In any event, twenty men in House A and twenty-six men in House B completed both questionnaires and designated two or more persons who had also completed the initial questionnaire. These numbers represented not only about half of the student residents in the house during the year but also about three quarters of those not going to an overseas campus during the year.

RESULTS

Because the samples were small we combined both houses for all predictions and analysis. Three groups were distinguished on the basis of the drug-use information given on the first and second round of questionnaires. One group, comprised of twenty-one students and, designated as the 00 (zero-zero) group, reports no use of any illicit-exotic substance either at the beginning of the school year or at the end of the year. The second group reports no illicit-exotic drug use at the beginning of the year but by year's end says members began drug use, usually marijuana but possibly LSD, peyote, and so on. There were fifteen students in that group which we designated as the 0+ or zero-plus group. The third group of ten students, designated as the ++ or plus-plus group, reports on the first round of questionnaires that members have had experience with illicit-exotic drugs and so necessarily have had experience on the second round as well. In October at the beginning of the year only 22 per cent of the students in the two houses had had experience with illicit-exotic drugs (a figure that corresponds quite closely to our overall sample of the same university taken during the fall quarter) but that by June 55 per cent had had such experience. It is in the comparison of the group of fifteen (33 per cent) who turned on with those twenty-one (45 per cent) who did not begin use that we are primarily concerned. We shall also present the findings on the plus-plus group in the expectation that as experienced users their scores on willingness and so on would be highest.

One assumes that what people say they are willing to do has something to do with what they in fact do. So it was that we expected that those who had turned on during the year would have had a greater expressed willingness to use drugs when asked at the beginning of the year than would those who had not turned on. Recall that our willingness scale was divided into two components, experimental and social; we expected that social willingness—which is the kind ordinarily involved in the taking of illicit-exotic drugs—would be more relevant than experimental willingness. The individual willingness score used was the average score (minimal 1, maximal 5) on each of the eighteen drugs on the scale (including the two fictitious drugs).

We find that the zero-zero group received an average experimental-willingness score of 2.13 and an average social-willingness score of 1.54. The zero-plus group received an average experimental-willingness score of 2.75 and a social-willingness score of 2.04. In contrast', the plus-plus group received an average experimental-willingness score of 3.22 and a social-willingness score of 2.34. Because the groups are small, these differences are not statistically significant; they are, however, in the predicted direction in that members of the zero-zero group consistently show the lowest willingness scores, both experimental and social, members of the zero-plus group show higher scores, and those of the plus-plus group show the highest scores. Note, however, that social willingness does not differ from experimental willingness. The willingness score of even the experienced subjects is itself moderate in the sense that an average score of 3 represents a self-rating as "neutral" on taking drugs, whereas an average score of 2 represents slight unwillingness. On the other hand, to gain an average score much higher than 3, subjects would have had to express high willingness to take drugs with known characteristics productive of physical dependency. It appears that overall willingness is too broad a measure to pick up eagerness to take particular drugs.

One can, for this reason, focus on willingness as expressed in regard to those drugs on the list that were exclusively illicit: psilocybin, marijuana, LSD, heroin, hashish, and peyote. There we find that the zero-zero group has an average experimental willingness of 2.3 and a social willingness of 1.4; the zero-plus group has an average experimental willingness of 2.6 and social willingness of 1.8, whereas the plus-plus group has an average experimental willingness of 3.4 and a social willingness of 2.7. Once again, the results are not significant although they are, as one would anticipate, in the expected direction. What stands out is that social willingness does not differ from experimental willingness in terms of discriminating among the groups and also that willingness limited to illicit-exotic drugs is no better a measure than the overall willingness scale.

We expected that students who turned on would have, at year's beginning, more friends who were themselves experienced drug users than those who did not. Our assumption was that the choice of friends experienced in illicit drugs was itself an indicator of acceptance if not interest in that kind of conduct and, secondly, that there would be more opportunity for drugs and social pressure to use them from such friends. The questionnaire called for subjects to estimate the number of drug-using friends they had on a four-point scale ranging from "none" (value = 0) through "very many" (value = 3 ) for each of three drugs: marijuana, LSD, and "other drug or medicine for kicks, to get high." The maximum score for any subject (very many friends using each) would be 9, the minimum 0 (no friends using any).

One finds that the zero-zero group has an average score for estimated friends' use of illicit drugs of 1.6, the zero-plus group has a score of 2.9, and the plus-plus group a score of 5. These differences are clear-cut and in the predicted direction, although again our small N means that they are not statistically significant.

We expected that the attitudes of each subject's own friends and leaders in the fraternity house would also be associated with his turning on or not. We assumed that when the friends and leaders nominated by him (the sociometric nomination called for them to name the five best friends within the living group and, separately, the five most respected individuals) were themselves more willing to take drugs, then these attitudes would support or encourage the subject to do so. Such attitudes were not likely to be separate from the drug experience of these friends and leaders. Our measure of attitudes here was the willingness score of those nominated as friends and respected leaders.

One finds that the zero-zero group nominate friends with average experimental willingness of 2.5 and social willingness of 1.8; the zero-plus students nominate friends with scores of 2.3 and 2.1; the plus-plus group nominate friends with scores of 2.8 and 2.1. With reference to respected leaders, the zero-zero group nominate students whose average social willingness is 2.5 and whose experimental willingness is 1.7; the zero-plus group nominate leaders with average scores of 2.7 and 1.9; the plus-plus students nominate leaders with average experimental-willingness scores of 2.8 and social-willingness scores of 2.0. Slight trends notwithstanding, it is clear that the willingness of selected friends and leaders, as reported by those friends and leaders, little distinguishes the students who are drug experienced, those who turn on, and those who stay aloof from illicit experimentation.

Subjects were asked for each of the eighteen drugs on the test list whether they had used it and, if so, how often. Replies were grouped with a range of scores from 1 for little use to a maximum of 4 for considerable (ten or more times) use. Total scores were for the sixteen real drugs on the list. We expected that students in the zero-plus group would have higher past-experience scores than those in the zero-zero group.

One finds that the average total scores for the zero-zero group are 4.5, for the zero-plus group, 6.9, and for the plus-plus group, 11.5. These are in the anticipated direction although, once again, they cannot achieve statistical significance.

If one excludes alcohol and tobacco, one finds that the average past-experience score for the zero-zero group is 1.1, for the zero-plus group, 2.0, and for the plus-plus group, 4.9. If one further excludes illicit-exotic experience, which can affect only the plus-plus group, the average score of that group drops to 1.8.

It is of interest to see what happens to these groups on the end-of-year retest. At that time the average past-experience score (excluding alcohol and tobacco) for the zero-zero group rises slightly to 1.2, for the zero-plus group it rises to 2.9, and for the plus-plus group it nearly doubles to 7.3. Clearly, the year of observation (or conceivable retest effects) has been an active one during which all students increased their drug experience. The already experienced students increased it most, followed by the newly turned-on students in second place. The implication is that once begun, use of illicit-exotic drugs among these fraternity students led to considerably more such use.

Illicit-exotic drugs can be obtained only through informal channels. We expected that the students in the turned-on group would have more experience using friends and acquaintances as drug (informal) sources than would the students not turning on. For the purpose of this comparison, sources were limited only to those druge (phenobarbital, benzedrine, seconal, Demerol, dexedrine, Miltown/Equanil, codeine, methedrine, and also alcohol and tobacco) which could be obtained either from doctors or from informal sources. We find the zero-zero group averages 1.8 drugs from informal sources and 1.2 from doctors; the zero-plus group averages 1.9 from informal sources and 1.7 from doctors; and the plus-plus group averages 2.7 drugs from informal sources and 1.1 from physicians; this latter group has an elevated informal-sources score because of members' experience with illicit-exotic drugs. The trend is as anticipated but is hardly dramatic.

Each student was asked six questions which touched upon his future plans, and these were condensed in scoring to specificity of plans with reference to future activities, relevance of present schooling, and family plans (marriage and children). We expected that the zero-zero group would show the greatest certainty in planning and the zero-plus group less.

We find, after assigning scores, that the zero-zero group averages 4.9 for future-activity planning compared with 4.1 (lower scores mean less certainty or specificity) for the zero-plus group, and 4.5 for the plus-plus group. Scores on school relevance for future activities are 4.8 for the zero-zero group, 3.8 for the zero-plus group, and 4.2 for the plus-plus group. Family-planning scores are 3.0 for the zero-zero group, 2.6 for the zero-plus group, and 2.8 for the plus-plus group. These are in the anticipated direction for the two prediction groups, although the slightly greater certainty scores for the students with high drug experience is not what we expected.

There was also an open-ended question under plans which asked what the student expected to be doing in five years if he followed the trends and interests of his present life. Replies were categorized into three groups: those with utopian or idealistic themes; those still "in process" or continued moratorium as, for example, staying in school or in military service; and those with settled Establishment expectations. We expected the zero-zero group to have the most certain and conventional plans. We find that in the zero-zero group half of the replies anticipate being settled in Establishment activities in five years, among the zero-plus group one third anticipate being settled in regular jobs, and among the plus-plus group only one tenth expect such routines. "Don't know" accounts for only 5 per cent of the zero-zero replies but for one third of the zero-plus replies and for one fifth of the plus-plus group. Idealistic-utopian expectations appear, in round one, most often in the plus-plus group.

Interestingly enough, on the second-round inquiry, at the end of the school year, shifts one might have expected did not materialize —that is, the zero-plus group had not become more idealistic-utopian. Indeed, the greatest shift had occurred among the zero-zero group, where three students (3/21) became more uncertain and, dropping their expectations of Establishment routine, entered the "don't-know" group while one in the "moratorium" group moved to an idealistic plan. What this suggests is that career uncertainty can increase during a school year without being associated with the onset of drug use. Even so, by the end of the year, both drug-using groups remained highest in career uncertainty and in moratorium-type expectations, although by year's end the zero-zero group was highest in plans for an idealistic career.

We find that among the fifteen expectations advanced, fourteen have proven to be in the anticipated direction with regard to the differences in the beginning of the year between those students who later begin to use drugs illicitly and those not initiating such drug use. The only- exception is a very slight reversal in the experimental willingness to take drugs on the part of nominated friends. In all fifteen cases, the plus-plus group have scores in the predicted direction vis-à-vis the zero-zero group, and in eleven cases these scores are also more extreme, as anticipated, than those of the zero-plus group. Individual tests of these differences fail to achieve significance because of the small N and the narrow range of scores involved; however, the probability of fourteen out of fifteen findings being in the predicted direction or, more extreme, of twenty-nine out of thirty using the plus-plus group as well, is significant—that is P < .05.

A prediction table: Aside from the small satisfaction in seeing that trends in scores were as expected, we wondered what our maximum success would have been in individual prediction if we had employed each of the major measures to construct a prediction table. To construct that table we did three things: first, we selected nine major variables to include; secondly, we revised scoring procedures so that the range of scores on each was about the same—that is, scores ranged from 0 to 5; and third, we summed each individual score on each of the nine variables and averaged them for each drug-use group. These scores were given equal weight and an optimal cutting point was established.

If a cutting point is established at between the average score of 2.2 and 2.3, we find that sixteen students in the zero-zero group are below it as are four students in the zero-plus group (and none in the plus-plus group). This means that if, at the beginning of the year, we had used these scores and predicted that all students with a score of 2.2 or less would not turn on and that all students with a score of 2.3 or more would either be turned on (or already be turned on) during that coming year, we would have had four false negatives and five false positives along with twenty-seven predictions. All told there would have been twenty-seven successes and nine errors compared with a chance expectancy—had we but flipped a coin—of 18/18. Our success rate of three out of four capitalizes on hindsight and on the remarkable set of natural events which led nearly half of the nondrug users in the house to become drug initiates during the year of observation.

The prediction table demonstrates that within homogeneous groups subject to similar environments, one can predict illicit-drug use with some accuracy simply by knowing the subjects' past drug experiences, past drug sources, reported willingness to take drugs, subjects' estimates of friends' use of illicit drugs, and willingness of nominated friends and leaders within the same living group to take drugs. By adding to drug use another variable, future plans, and constructing a new prediction table, it is possible to further increase this after-the-fact "prediction." The new cutting point appears between those with average scores of 2.4 and 2.5. Seventeen of the zero-zero group fall below the cutting point and three above; four zero-plus students fall below and twelve above. Furthermore, all plus-plus students fall above the 2.4 point. Thus, we would have twenty-nine successes and seven errors or a rate of success of 81 per cent compared with the chance expectation of 50 per cent.

The foregoing indicates what the next step would be, which would require the use of the same scales at the beginning of a study in a similar setting. Even were one lucky enough—if one calls it "lucky" to be on a campus where marijuana use doubles in a year—to have some students turn on and others not, it would be unrealistic to expect the same table to yield a rate of success of prediction of 81 per cent or the even higher rate of success which emerges if one wants the table simply to identify all those who by year's end will be illicit users. Taking the latter tack, which requires the plus-plus group be included in the prediction, leads to correct identifications of student drug use in 85 per cent of the cases. The reason that cross-validation efforts will show lesser success rests essentially on one's having maximized success after-the-fact on one sample; the next sample will be bound to differ in some ways even if it is drawn from similar environments. That sample will also be exposed to altered environmental forces as the expanding attraction of drugs affects campuses. The statistical problems in prediction and cross-validation were long ago discussed by Horst (1948).

SUMMARY

An exploratory study has shown that, as measured by a scale among (male) students, stated willingness to take drugs is significantly correlated with their response to social invitations to use illicit drugs. Two components, willingness to take drugs in experimental settings and willingness to take them in social settings, are also found to be significantly related, even when all drugs taken in the past are excluded froni the scales. Other correlations achieving significance are between social willingness and the number of friends estimated to be illicit-drug users and between general curiosity and social willingness.

Using the willingness scale, and an inquiry into future plans, plus sociometric ratings of friends and leaders, observations were then conducted in two fraternity houses over the course of the 1966-1967 school year. All residents were tested at the beginning of the year and again in late spring. They were classified at year's end as having used illicit drugs from the beginning of the year, as having turned on for the first time during that year, or as having never turned on to illicit drugs. At the beginning of the year, 22 per cent had had illicit-drug experience; by the end of the year 55 per cent had had such experience. Findings show that on fourteen out of fifteen variables based on the scales administered (willingness, experience, drug sources, friends' and leaders' actual use, and friends' and leaders' willingness and future plans) the group not turning on received lower scores than the group turning on; furthermore, on all fifteen variables, the group already experienced with illicit drugs received higher scores than the group not turning on, and on eleven variables received average scores higher than those turning on during the year of observation. Constructing a prediction table based on the average scores on these variables, we found that those turning on had higher scores on the average than those not doing so—to such an extent that, after the fact, there was a success rate of 81 per cent in identifying students as having turned on or not. Although future studies utilizing the same prediction table among similar samples cannot be expected to achieve the same success as an after-the-fact construction, we do conclude that among homogeneous student populations the prediction of who will turn on can be accomplished at a better-than-chance level by attending to the variables of drug experience and willingness among students and their friends.

1 We have reason to believe that females would show a lower correlation—our observations indicate they respond more in terms of who the offering person is and what their relationship to him is.

2 There were six questions asking about interests in seeing new places, meeting new people, trying new foods or cooking, participating in psychology experiments, in medical research experiments, and experiments to evaluate new drugs. Thus, one component of the curiosity scale was itself a drug-willingness item.

3 The interest-curiosity variable which also correlated was discarded because of the overlap between it and experimental willingness items.