APPENDIX G Reasons for Not Taking Heroin

In reviewing the two delinquency-orientation clusters and the drug-subculture cluster in High, we found that ten of the eleven items in the questionnaire which dealt with reasons for not taking heroin formed part of the drug-subculture cluster and that nine of them were included in one of the delinquency-orientation clusters; the eleventh item was, as already explained, not included in the cluster analysis. We now note that these ten items and an eleventh composite item, G 31-41 ("Agreeing to nine or more of the eleven arguments"), form prime clusters in each of the three neighborhoods. What does this clustering betoken?

One possibility is that these items tap a yes- (or no-) saying tendency in the subjects that functions independently of the content of the items. All are identical in form and call for a response of "yes" or "no." In favor of the possibility is the fact that, in all three neighborhoods, substantial proportions of the boys (35 per cent in Low, 46 per cent in Medium, and 55 per cent in High) said "yes" to nine or more of the eleven arguments. Against the possibility, however—and compellingly so in our estimation—may be cited the following facts: In the first place, the items associated with the prime clusters do not follow the same pattern. The associated clusters include two items in Low that do not call for a yes-no or agree-disagree response, one such item in Medium, and three in High. Even more to the point, each of the clusters includes items that, from the point of view of agreement, are scored in the direction opposite that of the prime cluster. There is one such item in Low, one in Medium, and eight in High. In addition, the Low cluster involves a compound item in which one of the three components has to be answered in the direction opposite that of the items in the prime cluster, although the other two are scored in the same direction. The Medium cluster also includes this compound item and another in which all three components have to be answered in the opposite direction. The High cluster includes the latter compound item.

In the second place, each of the arguments was paralleled in the questionnaire by a more general item which (except for the specific reference to heroin and a structural difference which will be noted below but which is irrelevant to the present discussion) is extremely similar to it in both form and content. The parallel items are the value items. Thus, Item G 35 ("People will look down on you") involves the value of achieving favorable recognition and is paralleled by F 30 ("To be very popular and have a lot of people look up to you"). Similarly, G 37 ("You'll lose your chances for a good steady income") is paralleled by F 35 ("To have a job you can count on and know that you can always get along"). There was also a discernible tendency to say "yes" to the value items.' The important point, however, is that, not only did the value items not cluster with the set of argument items, but they did not even form a single prime cluster of their own in any of the neighborhoods.a Moreover, the index of agreement with the arguments bore little relationship to the index of agreement with the value items. The tetrachoric correlation between the two indexes was .31 in Low, .39 in Medium, and .19 in High. The profile correlations were .21, .22, and .07.

The clustering of the argument items, then, cannot be accounted for in terms of a generalized tendency to agree or disagree. We may, therefore, look to the specific contents of the items in the augmented clusters for clues as to the character of the prime clusters.b
Some speculation may first be in order as to why the arguments did not divide themselves into the same kinds of differentiated clusters as did the value items. Apart from a major structural difference between the two sets of items,2 the clustering of the arguments was affected by whether the subject believed them. If the subject did not believe that taking heroin would have a particular consequence, then, regardless of his attitude toward the consequence, he would reject the corresponding argument. If he had a generalized disposition to disbelieve that evil consequences flow from the use of heroin, he would tend to reject all the arguments. Contrariwise, if the subject believed that a particular consequence would ensue, then, again regardless of the relative position of the entailed value in his hierarchy of values, the corresponding argument would tend to become a major deterrent. All the values involved in the arguments tend to be of considerable, if not of equal, importance to all people. If he had a disposition to believe that many evils flow from the use of heroin, the respondent would tend to accept all the arguments, including arguments the soundness of which he could not evaluate in any realistic fashion. The general disposition to accept or reject arguments might thus be indicative of a general attitude toward the use of heroin or even more general attitudes in which the attitude toward heroin was imbedded.

In light of these considerations, it is not surprising that one of the differences between the neighborhoods was in the number of items that clustered with the prime argument clusters. There were seven additional items in Low, eleven in Medium, and eighteen in High. That is, on the assumption that the clustering of the argument items reflected more general attitudes, thee latter were least pervasive in their implications in Low and most pervasive in High. Moreover, the particular items entering the augmented clusters supported the assumption, although the pattern was rather different in Medium than it was in Low and High.

In Medium, the cluster developing around the arguments included a complete prime cluster of what may be referred to as "high ideals." Specifically, the acceptance of the arguments went with F 27 ("Values intimate companionship"), F 32 ("Values workmanship and being good at sports"), F 34 ("Values being of service"), F 36 ("Values conformity and keeping up with people you like"), and F 39 ("Values refinement"). The rejection of the arguments, of course, went with the rejection of these values. In addition, accepting the arguments went with F 2, 16, and 21 ("Favorable to parents on all three items"); G 1 ("Taking heroin not a private affair"); G 2 ("Tell authorities about boy on block who uses heroin"); G 3, 4, and 5 ("Antidrugs on all three items"); G 6 ("Idea of smoking marijuana worst thing can think of"); and G 9 ("Heroin-users have fewer brains").

Although five of the eleven supplementary items also occurred in the Medium delinquency-orientation cluster3 and three of the five in the prime cluster as well as in the general delinquency-orientation cluster, the predominance of other items and the relative independence of the prime argument cluster from the lattere suggest that the delinquency and argument clusters are derived from relatively independent cultural currents. The inclusion of the prime-value-item cluster in the argument cluster suggests that the latter  derives from a cultural current concerned with what would be described, in middle-class society, as the finer things in life.

In both Low and High, we had similar value clusters, but these did not become involved in the argument clusters. In other words, the acceptance or rejection of the arguments tended to derive from cultural currents that were independent of those affecting the acceptance or rejection of the "high-ideals" value items, and, as we shall see in a moment, the tendency to accept or reject the arguments was not independent of the delinquency-orientation and drug-subculture currents. The Low argument cluster included none of the value items, and the High included only one, F 39 ("Values refinement"). All the seven supplementary items in the Low argument cluster were also in the neighborhood delinquency-orientation cluster—three in the prime cluster and two of the latter plus one other in the general delinquency-orientation cluster. Similarly, sixteen of the seventeen supplementary items in the High argument cluster were in the neighborhood drug-subculture cluster.4 The one item that seemed out of place was F 39 ("Values refinement").

A reasonable interpretation of the patterns is that the clustering of the argument items was influenced in Low by a generalized attitude or set of attitudes generated by contact with the cultural currents that produced the delinquent subculture and, in High, by contacts with the cultural currents that shaped the two delinquency-orientation and the drug-subculture clusters.5

Two other aspects of the responses to the argument items merit some attention—an apparent paradox and the comparative rank order of the arguments.

First, the paradox. Fifty-five per cent of the boys in High accepted eight of the ten arguments in the prime cluster, as compared to 47 per cent in Medium and 35 per cent in Low.6 More than 25 per cent accepted all ten, as compared to 16 per cent in Medium and 14 per cent in Low. On the face of it, these percentages diminish in the wrong direction. That is, one might expect that, if anything, there would be the least acceptance of the arguments in the area of highest drug use, rather than the opposite. The same paradox appears at the low end of the scale: only 10 per cent of the boys in High rejected seven or more of the ten arguments, as compared to 12 per cent in Medium and 18 per cent in Low. Correspondingly, the average number of arguments accepted was greatest in High and smallest in Low.

The apparent paradox, however, presumably stems from our presuppositions, rather than inhering in the facts themselves. It seems reasonable to assume that, the more pervasive the illicit use of narcotics, the more pervasive is the evidence that there are many evil consequences associated with the practice. Presumably, the youngsters with whom we were dealing were not immune to the testimony of their direct or vicarious experience.' Hence, it was not at all paradoxical that the youngsters in High should accept more evils as associated with heroin use, and, as already noted, the arguments were such that accepting the consequence made it difficult to reject a given argument.
Where, then, is the paradox? It is not in the tendency to accept a greater number of arguments in High, but in the fact that the higher level of acceptance was not sufficient to overcome the attractiveness to many of the youngsters of the drug-using and delinquent subcultures. This, however, places the paradox in our naive assumptions about human nature, viz., in the assumption that agreeing that some practice is dangerous and that accepting the appreciation of the danger is a deterrent sufficient to shape one's attitudes toward the practice. Suppose, for instance, we were to ask adolescents with the habit of cigarette-smoking whether they believed that cigarettes cause cancer and found that most agreed that they probably do. Suppose, further, that we asked them whether the relationship between smoking and cancer "is one of the main reasons that would keep you from smoking" and found that most said "yes." To many people, such an outcome (which, in the opinion of the writers, is highly probable) would be incomprehensible. The point is, however, that how one responds to direct confrontation with an issue bears no necessary relationship to whether one normally and spontaneously confronts that issue and, if one does, to one's capacity to segregate that response from other relevant attitudes and practices. Moreover, one of our cigarette-smoking adolescents might well say to us: "Of course, the danger of cancer is one of the main reasons that would keep me from smoking, but who said that the danger of cancer tells the whole story?"

In this light, the concurrence of a greater readiness to agree that various dire consequences of heroin use are major deterrents with an attraction to the drug-using subculture and a sharing of its attitudes does not seem so paradoxical. At the same time, we should not forget that, relative to the other youngsters in High, those who tended to share the attitudes of the drug-using subculture and, to a somewhat lesser extent, those who tended to share the pessimistic version of the delinquent subculture were the ones most likely to reject the arguments.7

The second additional aspect of the data on the arguments that merits attention has to do with the rank orders of the arguments in the three neighborhoods. In each neighborhood, we could determine the argument with the highest percentage of acceptance, the argument with the second highest percentage, and so on. The resulting rank orders were quite similar in the three neighborhoodsd despite the differences in the tendency to give blanket assent to the arguments and the differences in the absolute levels of the individual arguments.

In all three neighborhoods, the most agreed-to reason for abstaining from heroin was G 41 ("Health will be ruined and life full of worries and troubles"). Other relatively strong arguments in all three neighborhoods were G 39 ("Become too different from other fellows"), G 37 ("Lose chances for good steady income"), and G 40 ("Won't be able to work well or be good at sports"). Relatively weak arguments were G 33 ("Kick wears off, but you still have to go on using") and G 35 ("People will look down on you"). The argument which showed the most marked change in rank was G 31 ("Become a slave to heroin"), which held second place in Low, third place in Medium, but ranked 8.5e in High. No other argument approached this degree of variation in its rank positions!

To summarize, in all three neighborhoods, there was evidence of a generalized tendency to accept or reject arguments against taking heroin, over and above the specific reactions to particular arguments. This tendency seemed to be part of larger constellations of attitudes that included aspects of the delinquent and, in the case of High, drug-involved subcultures of these neighborhoods. It was not, however, a direct expression of these subcultures, since there were many aspects of the latter not associated with the argument cluster. Despite the generalized tendency, the arguments formed a fairly stable hierarchic order of appeal in the three neighborhoods, the threat to health and the incidence of "worries and troubles" evoking the most nearly universal assent. The deterioration of capacity to work well and to participate effectively in active sports and the fact of tolerance (diminished reaction to drugs) along with continued addiction were reasons with relatively low appeal. The danger of addiction itself constituted a relatively strong argument in Low and in Medium, but it was one of the least persuasive in High. The generalized tendency to accept arguments was strongest in High, possibly as a consequence of a counter—to—drug-using subculture generated by a more pervasive concern with issues of drug use. A point of caution worth bearing in mind is that agreeing that something is one of the most potent deterrents is not the same as accepting that something as a sufficient deterrent.

STATISTICAL FOOTNOTES

a Three of the value items were members of the same prime clusters in each of the three neighborhoods. These were: F 32 ("Values workmanship"), F 34 ("Values being of service"), and F 39 ("Values refinement"). Three others were members of the same prime clusters, but other ones than the first three. These were: F 28 ("Values constant novelty and excitement"), F 33 ("Values freedom from restraint"), and F 38 ("Values thrills and chances").
The first trio constituted a prime cluster in High; in Medium, it constituted a prime cluster together with F 27 ("Values intimate companionship") and F 36 ("Values conformity and keeping up with people you like"); in Low, the last five items, together with F 40 ("Values freedom from sickness and yam") made up a prime cluster. Acceptance of one of the arguments was part of the augmented cluster in Low, of six of the arguments in Medium, and of four in High. G 34 went with the Low and High clusters; G 33 and G 40 with the Medium and High; G 32, G 37, G 38, and G 39 with the Medium; and G 35 with the High.
With reference to the second trio of value items, it constituted a prime cluster in Medium; together with F 29 ("Values power"), F 30 ("Values being popular and looked up to"), and F 31 ("Values being able to take things easy and not having to work hard"), it made up a prime cluster in Low; and, with the last three items, F 36 ("Values conformity and keeping up with people you like"), and F 37 ("Values being strong and manly"), it made up a prime cluster in High. Note that F 36 was associated in prime clusters with the first trio in Low and Medium and with the second trio in High. F 37 was a relatively strong member of both augmented value clusters in Low and High and of the first value cluster in Medium; it did not occur at all in the augmented second cluster of the latter neighborhood. None of the argument items were associated with any of the second value clusters. Presumably, quite different aspects of the meaning of F 36 and F 37 were tapped in the various cluster contexts.
The average profile correlation of the two prime value clusters in Low was —.02, and the average tetrachoric, .02; the corresponding statistics for Medium were .20 and .13; for High, .55 and .28. Although the profile correlation in the last-named neighborhood was high enough, by our criteria, for the two prime clusters to be joined in one cluster, the between-cluster statistics may be compared to the within-cluster statistics for purposes of perspective. In High, each item of the first prime cluster had an average profile correlation with the others of .74, and the lowest average tetrachoric was .40; for the second prime cluster, the lowest average profile correlation of any item with the other was .70, and the lowest average tetrachoric was .31. Scores based on the two prime value clusters (using only the items common to the three neighborhoods in each prime cluster) correlated to the extent of .20 in Low, .14 in Medium, and .23 in High. (These are regular Pearsonian correlations, not tetrachorics.) When the effect of the tendency to agree with the value items is discounted (by a statistical technique known as partial correlation), these correlations become, respectively, —.48, —.46, and —.45. That is, basically, the two sets of values tend to be mutually contradictory.
There may be some interest in the distribution of scores on the common items of the prime value clusters. The strongest appeal of the first set was Medium, where 56 per cent of the boys said "yes" to all three items; the weakest, in Low, where 36 per cent accepted all three. One subgroup in Medium stood out, the white parochial-school boys, with 71 per cent getting the maximum score. In High, 52 per cent scored three. On the second trio, there was little variation in the scores, except for the adjustment-class boys in High. Over all, half the boys scored two or more; but, in the adjustment classes, this figure ran to 62 per cent.

b This is not a contradiction of what was said earlier concerning the importance of looking at the prime clusters before attempting to interpret the augmented clusters. In the present instance, however, we are dealing with three identical prime clusters (with the minor exception that G 2 joins the prime cluster of arguments in Medium); the only clues as to possible differential meanings of the prime clusters come from the items in the augmented clusters. Meaning is, of course, an interaction process.

c The average profile correlation in Medium between the items in the prime argument cluster and those in the prime delinquency-orientation cluster was —.33; the average tetrachoric, .02.

d The coefficient of concordance was .79. This coefficient is an index of the correspondence between several sets of rankings. A coefficient of 1.00 would mean that all the rankings were identical.

e This means that this argument was tied with another in percentage of acceptance, the two arguments being in eighth and ninth place; in such a case, the items involved in a tie are assigned the average of the ranks for which a tie exists. Note that, in the present instance, there were two items with lower percentages of acceptance (ranks 10 and 11) and seven arguments with higher percentages of agreement (ranks 1 through 7); this left ranks 8 and 9 for the two tied items, and each item got the average of 8 and 9, viz., 8.5.
The next most discrepant ranks were for G 34 (with ranks of 7.5, 4.5, and 8.5 in Low, Medium, and High) and G 38 (with ranks of 7.5, 9, and 5).

1 In Low, 47 per cent of the boys said "yes" to nine or more of the fourteen value items, 52 per cent in Medium, and 48 per cent in High.

2 It is to be noted that all of the arguments involved the negative of values. For obvious reasons, we could not parallel the value items with arguments for the use of heroin. In principle, however, we might have done so. For instance, we might have included the following: "Suppose you believed the following statements to be true. Which of them would most influence you to try taking heroin?. . :Taking heroin makes a person feel strong and manly.' . . . Would this be one of the things that would be most likely to lead you to try taking heroin?" On a priori grounds, we would expect that such arguments would indeed show a structure similar to that obtained with the value items. The point is that asserting a positive puts the content of the assertion into the foreground; in the language of Gestalt psychology, such an assertion has figural quality. The subject has to indicate whether the given figure is also present in his life space. The assertion of a negative, however, simply tends to leave one with no content. To say that something is or will become absent does not say anything about what will happen at the site of the void. The subject has to evaluate how much of a difference the absence of a figure will make in his life space. Absences, we would suggest, are much less finely differentiated from one another than presences.
Granted, however, that this mechanism for the lesser differentiation of the arguments is operative and contributes to the over-all effect, it seems to us that the explanation advanced in the text must be primary not only because of its prima-facie plausibility, but because it also helps to account for the neighborhood differences in the complete argument clusters.

3 In referring to these items before, the paraphrases of the items were oriented in the opposite direction. Here, we are focusing on the acceptance of the arguments and have oriented the paraphrases accordingly.

4 Nine of the sixteen items in the general delinquency-orientation cluster were included; these nine items, of course, occurred in both the neighborhood delinquency-orientation clusters. Three other items, not in the general delinquency-orientation cluster occurred in both the neighborhood delinquency-orientation clusters. Three other items occurred in the second neighborhood delinquency-orientation cluster,. In all, fifteen of the eighteen supplementary items were found in the latter cluster. Only one of the supplementary items, G 12 ("Heroin-users get fewer kicks out of life") occurred exclusively in the drug-subculture cluster, and this was one of the weakest items in the latter set, with an average profile correlation of .43 and an average tetrachoric of .11 with the items of the prime cluster. It is, therefore, something of a tossup whether to describe the supplementary items as being drawn mainly from the drug-subculture cluster, from the second delinquency-orientation cluster, or from contact with the cultural currents that established the two delinquency-orientation and the drug-subculture clusters.

5 The full set of supplementary items in Low consisted of F 2, F 16, F 21, F 17, F 24, G 6, G 7, G 11, and G 47. In High: F 4, F 6, F 7-13-26, F 22,
F 23, F 24, F 25, G 3-4-5, G 8, F 5, G 2, G 47, G 6, G 7, G 12, F 39, and
G 17-25. In apparent contradiction to this interpretation, one may be struck by the relative paucity of image-of-the-heroin-user items in the clusters; three of the most relevant ones, however, were not included in the cluster analysis, for reasons already indicated. A fourth which seems directly relevant is G 13 ("Users have fewer close friends they can trust"). Logically, this seems to parallel G 36 ("Be alone, won't have real friends"). G 13, however, may be taken in the sense of an initial personal deficiency rather than as a consequence; e.g., it clusters with G 9 ("Users have fewer brains").

6 We have already mentioned corresponding figures when the eleventh argument is included. They are virtually identical (55 per cent in High accepted nine of the eleven, 46 per cent in Medium, and 35 per cent in Low) because of the very high percentages of acceptance of the eleventh argument.

7 Although we have noted that the major determinant of the acceptance of the arguments in Medium was associated with a cultural current concerned with the "finer" things in life, it is also true that those who shared the attitudes of the delinquent subculture in this neighborhood tended to be the ones most likely to reject at least some of the arguments. It may be recalled that three of the argument items were part of the delinquency-orientation cluster in Medium: G 32 ("Hurt people close to you"), G 36 ("Not have real friends") and G 39 ("Become too different"). In effect, we seem to have here a convergence of two cultural currents on these specific items: the negativistic individualism of the neighborhood's delinquent subculture, which would tend to generate an attitude that these cannot be good reasons for not doing something, on the one hand, and the concern with the finer things in life, with its high valuation of intimate companionship, doing things for others without regard to personal returns, and conformity, on the other.