8/8/2019 Ginsberg 1978

1/57

This document contains text automatically extracted from a PDF or image file. Formatting may havebeen lost and not all text may have been recognized.

To remove this note, right-click and select "Delete table".

8/8/2019 Ginsberg 1978

2/57

8/8/2019 Ginsberg 1978

3/57

and cross-sectional which to have which interpretations However, marijuana quences more use

public use sectional debate hampered tests 1973; Most use has of Drug reported of Groves,

confident policy temporally are marijuana marijuana. of past theories come of by survey use.use Abuse, these subject one its studies 1972; regarding a correlates 1974). represent data (cf.

heightened lack use. cannot issues designed conclusions have research 1972;

However, to order Abelson Dembo whether of of Longitudinal Along various why the adequate

employed

continues, of causes conclude variables to possession marijuana with concern et people designs

correlates et constructive

account theoretical

al., regarding

al., or increased

empirical

studies

1976).

conse-

1972,

allow

to come

from

over

anduse

and

be

for

of

Supported in part by Public Health Service

Grant MH 20708 (National Institute of Mental

Health) and by a grant from the Robert Wood

Johnson Foundation. We are grateful to Howard

Johnson, langer,

David Mechanic and two

anonynious reviewers for helpful comments on ear-

lier drafts of this paper.

use. The implications for reinterpretation of previous data are discussed.

markedly decade, Index,

8/8/2019 Ginsberg 1978

4/57

(cf. TheNational use especially in of the Commission

marijuana United among States college has in increased

Opinion

the students

past

use at follow-up is not related to

initial lack of commitment to conventional institutions, directly or indirectly, when statistical

controls for other theoretical variables are applied. In addition, psychological distress at

follow-up is associated with

te empirical tests which assess the relative independent predictive

power of theories designed to account for

is

hampered by a lack of

use. This paper presents a longitudinal

study of these of theories.

college student

with prior use which studies, allows orientation assessment toward of a the marijuana-using relative utility of reference

certain

group is the most substantial predictor of

use in this study. Contrary to earlier

conclusions based solely on cross-sectional data,

use, when other spurious effects

are removed. Finally, degree of involvement in conventional activities is not related to

Constructive debate over public policy concerning the use and possession of

COMPETING THEORIES OF MARIJUANA USE:

A LONGITUDINAL STUDY*

I

RVING

J. G

INSBERG

J

AMES

R. GREENLEY

Journal of Health and Social Behavior 1978, Vol. 19

levels of initial

of Wisconsin

8/8/2019 Ginsberg 1978

5/57

school-aged males from both personality

and perceived environment variables

22

what this useful critically adequacy investigated of environmental, use theories longitudinal lege

between multiple students However, dress marijuana regression of perceived example, cients

ceived ables

Sadava,

significant ing 1975; 1975; Several marijuana demographic, onset paper (Kandel students, may

Lucas, were Jessor predictors the environment onset marijuana correlation examine among of

of we or prior be analyses few use. environment study relative multiple the reported marijuana

1973; several from et a and present of and variables report 1973b; cause longitudinal extent

longitudinal al., Jessor high of of marijuana Faust, sets the Smith behavioral use

19733 predicting and personality,subsequent high theories.

coefficients for utility longitudinal correlation versus school-aged step-wise to of relative

respectively). et Mellinger use personality the and which 1975; school personality variables. al.

in Jessor, among

enables use studies an relationships

of studies Fogg, predicting

factors initial predictive

marijuana

effect. a Kaplan,

and and multiple

various

data predict-

coeffi-

variety

1976).

males,

et 1975;

us have

vari-in per-

The

col-

and

For

al.,

8/8/2019 Ginsberg 1978

6/57

ad-

use

are

on

In

to

a

and

8/8/2019 Ginsberg 1978

7/57

8/8/2019 Ginsberg 1978

8/57

8/8/2019 Ginsberg 1978

9/57

suggesting that these sets of variables may

not be independent predictors of

marijuana use. No competitive tests usingstatistical controls were reported to eval-

uate the relative independent predictive

power of these two theoretically impor-

tant sets of variables or to assess the rela-

tive predictive utility of specific variables

within these groups.

Sadava (1973b) reports a six-month lon-

gitudinal study of college students using

various measures to predict future levels

of marijuana use. Frequency of marijuana

use was regressed on the predictor vari-

ables simultaneously; social support for

marijuana use, belief in Internal vs. Ex-

ternal control, and the experience of nega-

tive functions of marijuana use remained

significantly related to marijuana use in-

dependent of other factors. These three

variables predicted marijuana use with

multiple R =

data are difficult to

interpret because in predicting marijuana

use from other variables, the possible ef-

fect of the initial level of marijuana use is

not considered. Thus, relationships be-

tween predictors and subsequent fre-

quency of marijuana use may be due

solely to associations of these predictors

with initial levels of use and stability of

use over time (cf. Heise, 1970). Studies

such as Jessor et al.

which predict the onset of marijuana use

beginning with a non-user group, escape.

8/8/2019 Ginsberg 1978

10/57

this problem; however, these studies fail

to differentiate adequately the relative in-

dependent predictive power of prominent

competing theories.

These longitudinal, and other

Clearly, social support for marijuana

use was the strongest predictor of fre-

quency of marijuana use in this study. In

addition, Sadavas data demonstrate that

variables predicting the use of marijuana

over time may themselves be predicted

from prior levels of marijuana use, and

several other studies report changes in

these variables concomitant with changesin levels of marijuana use (Jessor et al.,

1973; Jessor, 1976; Sadava, 1973a). How-

ever, Sadavas

Lucas et al.

THEORIES OF MARIJUANA USE 23

combined was not much larger

while social support

alone predicted marijuana use with R =

sectional, studies suggest a wide variety of

and Sadava

Jessor

=

theories examined Butler, present certain parental and (Jessor sons tion, quality dressed ing use

Johnson,

8/8/2019 Ginsberg 1978

11/57

the (Single 1973). studies and and 1973b; lems al.,

(Kandel progression the Fejer, from validity over for some in Certain 1976; of et et specific

study, Kaplan,

use issues of legal drug-use of using and marijuana al.,

al., the 1970; time 1972; using family marijuana longitudinal of of Lettieri, Faust,

1973; and changing of other marijuana including marijuana; from longitudinal personality

sequences (Kandel Jessor, these Kandel,

surveys data illegal relationships Sadava, methodological to than legal 1975; use self-reported

other theories rates reported studies 1976). drug those (Dvorak,

factors to (Gorsuch (Haberman et the these Braucht and characteristics

drug 1971,

illegal of illegal al., use reliability

cannot involving

stages marijuana

have In and includepredict-

(Blum,

surveys

in (Smart

1976;

et addi-

drugs

drugs

prob-

rea-and

the

the

al.,

ad-

be

in

et

and drugs issues ins they present et will by al., are study.

returning not 1974, important constitute 1975). Vietnam in the theirAlthough veterans focus own

the for use right,

these

(Rob-

the

8/8/2019 Ginsberg 1978

12/57

ofTHEORIES OF MARIJUANA USE

ence contrasted person emulate group of Four reference group is may theories a others in group

use theory, this for marijuana of who a self-evaluation

report. person marijuana which use The

takes it. in suggests an A use as reference

attempt is will a that refer-

frame

be

to

a

1957; ence conduct reference ally, peer ord use the guise searchers sure; occurs, influence and

groups groups learning group of Shibutani, and conformity groups have to on at belief is

to which least of a college suggested smoke 1967). for source peers, for in that a to an part,campuses person peer properly, A of whether individual. person, that positive standards as group

belongs Many a marijuana

result (cf.

perceive

such in refer-

pres-

Usu-

the

are

re-

as

of

of

and Christie,

adaptation to marijuana use

8/8/2019 Ginsberg 1978

13/57

8/8/2019 Ginsberg 1978

14/57

8/8/2019 Ginsberg 1978

15/57

1972; Sadava, 1973a; Jessor et al., 1973;Jessor, 1976; Kandel, 1974; Thomas et al.,

1975; Linn, 1971; Lavenhar et al., 1972;

Gorsuch and Butler, 1976; Lettieri, 1975).

These arguments are central to what has

been called peer group theory, al-

though different elaborations of peer

group theory contain various additional

arguments (cf. NCMDA, 1972). Ones

peer group is always a group one is a

member of, although ones reference

group may or may not be a group to-which

one belongs. For example, a person may

have a positive reference group whose

standards he emulates and yet not be a

member of it. Thus, the reference group

concept is a more general concept than is

peer group, and similarly, reference group

theory is the more general theory.

Reference group processes are related

in complex ways to differential associa-

tion theory (Sutherland and Cressey,1966;

subculture (Lindesmith, 1968). For exam-

ple, greater exposure to a marijuana-using

subculture may lead to increased identifi-

cation with marijuana users as a reference

group. Thus, identification with marijuana

users as a reference group may constitute

a crucial link between exposure to a

marijuana-using subculture and the use ofmarijuana. In addition, having marijuana

users as a positive reference group may

lead to increased association with

marijuana users as well as to marijuana

use. Our focus will be on reference group

processes because these processes are

often thought to provide the causal mech-

8/8/2019 Ginsberg 1978

16/57

anism linking differential association, e.g.,

exposure to a marijuana using subculture,

with specific behaviors such as marijuana

use.

Commitment theory or control theory is

the second explanation for marijuana useexamined here. Commitment theory

suggests that deviant behavior (as

marijuana use is here conceptualized), will

be more likely the weaker are a persons

attachment and adherence to the norms

and values of the society in general

24 JOURNAL OF HEALTH AND SOCI

the effects, and define them as pleasur-

able; or as simply due to social modelingand imitation (Johnson, 1973; Becker,

1953;

1972) and the notion of a drug

1968; Goode, 1970;

AL BEHAVIOR

ment is societal The third theory, which we will call

stress theory, attributes marijuana use to a

persons desire to escape from personal or

psychological problems. Proponents of

this view cite studies showing that fre-

quent drug users score higher on problem

indices or measures of psychological dis-

tress than nonusers (cf. Steffenhagen et

al., 1972; Mellinger et al., 1975; Beaubrun

and Knight, 1973; Cross and Davis, 1972;

Ausubel, 1970). However, stress theory

does not necessarily suggest that themarijuana user is mentally ill; psychiatric

disorder may be seen as its own type of

often escape. viewed Rather, as marijuana a use is more

coping response in

which a person seeks a physiological or

recreational fears, anxieties, refuge boredom, from mild or depression.

to serious

8/8/2019 Ginsberg 1978

17/57

A fourth theory attributes non-deviant

behavior to the fact that a person may be

simply too busy doing conventional things

to find time to engage in deviant be-

havior

defined as support of majorinstitutions, institutionalized

goals, and institutionalized forms of be-

havior as they now exist. This support of

conventional several ways, activity as

may be (197 generated 1: 16-26)

in

argues, including beliefs, attachment to

others, and commitment to conventional

activities because deviant ones appear

less beneficial than conventional ones. In

this paper we make no distinctions among

these types of bonds to non-deviant be-

havior patterns. Adherence to traditional

religious, economic, and political values is

often used as a measure of commitment.

Johnson

and

Gergen et al. (1972) all incorporate varia-

tions of commitment theory into their ex-

planations of marijuana use.

was common in

juvenile delinquency prevention, where

for instance playgrounds and other recrea-

tional possibilities were provided to keep

adolescents busy and thus out of trouble.

With regard to college life, involvement

theory suggests that certain students may

spend so much time at their studies,

ton

This argu-

ment, which we will call involvement

theory following

(Johnson 1973;

8/8/2019 Ginsberg 1978

18/57

(Hirschi,

Knight et al.

1971).

8/8/2019 Ginsberg 1978

19/57

8/8/2019 Ginsberg 1978

20/57

8/8/2019 Ginsberg 1978

21/57

THEORIES OF MARIJUANA USE 25

ployment, or conventional student organ-

izational activities that they have neither

the time nor energy to use marijuana.

The relationships between reference

group processes, commitment, involve-

ment, and psychological stress have been

incorporated into more complex

arguments. For instance, Johnson (1973)

argues that encouragement from ones

peer group, as a reference group, is a

major motivating factor in marijuana use

and that lack of commitment to major

societal values and institutions predis-poses one to associate with peer groups

likely to be using drugs. However, data

are largely lacking allowing assessment of

these more complex explanations.

common among marijuana users: males,

persons

18 to 21, unmarried, middle

or upper class, urban, Jewish, and atheists

(NCMDA, 1972; Steffenhagen et al., 1971;

Gergen et al., 1972; Dvorak, 1972). How-

ever, other studies have found no rela-

tionship between some of these char-

acteristics and marijuana use, and several

suggest that different factors may be re-

lated to marijuana use among high school,

college, and post-collegiate age groups (cf

l

et al., 1973; Henley and Adams,

1973). Socio-demographic factors may be

variously related to or predisposing

toward lack of commitment to traditional

societal values, adoption of a drug-using

reference group, less involvement in con-

ventional activities, or psychological dis-

8/8/2019 Ginsberg 1978

22/57

tress. However, available research is in-

adequate to allow precise specification of

Some of the factors and processes lead-

ing to marijuana use are probably further

related through complex feedback

the

greater influence over time.

A number of studies have reported that

persons with the following

demographic characteristics are more

anisms and reciprocal causation (Goode,

For example, one question con-

cerns the extent to which having a

marijuana-using reference group leads to

marijuana use versus the extent to which

using marijuana leads one to adopt a

marijuana-using reference group. Using

the longitudinal data in this study we will

attempt to assess the extent of occurrenceof these processes, and which has

the manner in which these factors are re-

lated to marijuana use.

This paper uses longitudinal data to

contrast commitment, reference

involvement, and stress theory as expla-

nations for marijuana use. Other theories

exist but these four are among the most

widely known and best supported by

available data (cf. Cohen,

METHODS

A self-administered questionnaire was

mailed to a random sample of all students

8/8/2019 Ginsberg 1978

23/57

registered at the University of

Wisconsin-Madison in November, 197 1.

With repeated follow-ups an 82 percent

response rate was obtained, represented

by 1502 respondents. The 3 19 respondents

who were freshmen or sophomores at thetime of the original survey and who were

still registered students in January of 1974

were mailed a follow-up questionnaire and

86 percent responded, yielding a sample

with N = 274. A comparison of those stu-

dents who remained enrolled with those

who left the University during the interim

indicated that those 160 freshmen and

sophomores who left the University dur-

ing the two-year follow-up period were

significantly heavier marijuana users than

their 319 classmates who remained

enrolled.

In addition, respondents to the

= up survey

274) were compared with

enrolled nonrespondents

using

information from the original survey. Sig-

nificantly more respondents (8 1 percent)

than nonrespondents (58 percent) were 19

years of age or younger at the time of the

original survey. Respondents were also

significantly less oriented to the student

counter-culture, according to an index of

attitudes elsewhere reported (Greenley more and Mechanic,

extensively

1976). No other significant differences be-

tween these two groups were observed on

any other variables, including measures ofpsychological distress and marijuana use.

Each respondent was asked how often

in the past twelve months he or she used

marijuana. Response categories were

never,

8/8/2019 Ginsberg 1978

24/57

once or twice, from 3 to 12

=

8/8/2019 Ginsberg 1978

25/57

8/8/2019 Ginsberg 1978

26/57

8/8/2019 Ginsberg 1978

27/57

dents who are admired and identified with

need not be close friends or evan casual

associates of the admirer. However, ones

peers usually constitute ones reference

group

we focused on political,

religious, and economic institutions . From

a large number of available items in our

questionnaire relevant to these institu-

tions, with the aid of factor-analytic tech-

niques a particular item was chosen to

represent commitment to each of these

institutions. This procedure resulted in the

choice of items which we feel best repre-

sent commitment to institutions as meas-

ured in this study. The question regarding

religion was, how frequently do you

currently participate in religious serv-

ices?, with six response categories

supplied ranging from more than once a

suggesting that this measure may also be

highly related

Our measure of commitment is an

assessment of support by the student for

major societal institutions. Following

conceptual and empirical work by

and shorter than the time between because our a period

original

and follow-up survey was needed to allow

investigation of causal relationships

over time. Studies of response bias in

self-reported drug use suggest that our

self-administered questionnaire method

8/8/2019 Ginsberg 1978

28/57

probably does not lead to substantial dis-

tortions of the data (Luetgert and

Armstrong, 1973;

et al., 1972;

Whitehead 1975).

and Smart, 1972; Single et al.,a

To establish whether marijuan.a use was

characteristic of a students reference

group, respondents were asked to what

extent uses marijuana described stu-

dents they admire and identify with.

Five response categories were supplied

ranging from very characteristic to

very uncharacteristic. Obviously,

and more than 12 times. Thus,

our measure of marijuana use enables us

crudely to examine frequency of reported

use as well as use versus non-use. A par-

ticular twelve-month period was asked

about to avoid the

inherent in

interpreting lifetime use data (Josephson,

26 JOURNAL OF HEALTH

times,

and

certain peer group

Johnson

and

AND SOCIAL BEHAVIOR

week to never. The politically rele-

vant question was, how would you

classify your political beliefs? Seven re-

sponse categories were supplied, ranging

from very conservative to very radi-

8/8/2019 Ginsberg 1978

29/57

cal. The third question asked respon-

dents to indicate on a five-point scale from

strongly agree to strongly disagree

how they felt about the statement, Be-

coming a success is a matter of hard

work.Standardized scores on each of

these variables were added into an index

of commitment to conventional institu-

tions, where greater commitment was

indicated by agreement with the statement

on hard work, frequent participation in

religious services, and conservative polit-

ical beliefs. Because the statement con-

cerning hard work was not included in the

follow-up survey, our measure of com-

mitment to conventional institutions at

follow-up consists of an index of the other

two items constructed in the same man-

ner. As a further check on this variable,

we constructed several other measures of

commitment, largely drawing on a series

of other relevant but less preferable items,

and found that use of none of these other

measures led us to any different conclu-

sions.

In order to examine stress theory, a

number of measures of psychological dis-

tress and personal problems were exam-

ined. Some incorporated serious psychiat-

ric symptoms, some took into account the

seriousness of symptoms, and some dealt

with problems in various areas of life

(e.g., school, family, and work). All of our

measures of psychological stress eventu-

ally led us to the same conclusions. Theindex reported here is a summed score of

27 reported symptoms, each weighted by

the respondents assessment on a

point scale of how serious this symptom

was. Such commonly used symptom

8/8/2019 Ginsberg 1978

30/57

checklists contain psychophysiological

and psychological symptoms and have

been appropriately criticized by the

Dohrenwends

and Seiler (1973). Items for the

index were drawn from the Langner

Item Screening Score and scales being de-

veloped by Dohrenwend and his col-

leagues (Langner, 1962; Dohrenwend et

al.,

Items focus on common

Phillips and Clancey

8/8/2019 Ginsberg 1978

31/57

8/8/2019 Ginsberg 1978

32/57

8/8/2019 Ginsberg 1978

33/57

THEORIES OF MARIJUANA USE 27

week lege organizations dramatics, week which organized and was hours employed hours lems

sion, structed severe student A seeing measure and such a problems student to loneliness, reflect

things as of general was involvement such the others participating as number as anxiety, well do

hearing not.

of as was hours on depres-

in

voices

more

con-

per

be The marijuana ling necessary may and (X) spurious 1970).

1. A Controls used for measured effect follow-up considered multiple courses, be on was the

he the reported to associated relationship course was student campus use of added sports,

because per for effect examine Time regression employment, registered. marijuana will engaged

marijuana week at to work he the of indicated and Time activities spend 1 be with initial

marijuana our or number factors between and for so estimated, she path 1, longitudinal use in

another forth. To three every use the marijuana resulting were publications,

was at per and on analysis these of use number (cf. variable Time A hours follow-up

hours credit gainfullyweek.

campus

spent variable

control-

at student

Heise,

course

1 Time

data.

in willuse

are

per

the

for

of

8/8/2019 Ginsberg 1978

34/57

in

X

aRESULTS

juniors cent; but portion being over cent comparable times, never, original heavier freshmen

than from used slightly marijuana students 1972 In statistically said marijuana 12 this 3 academicthe once registered 36 32 and of to users underestimated times, sample reported they percent;

percent. 12 and heavy sample use two-year seniors percentages or at times, once never is

sophomores twice,. year, 20 significantly the students marijuana of common using 3 The in

percent of follow-up or students 9 to used 25 study the percent college twice, were: 12 14

proportion because percent marijuana on spring marijuana. times, and percent; users reported in

period. this greater and 12 is who said the

students,

increases

of probably

of or a campus

ceased

17 46 of 1974,

were

small

these

more

they

more

pro-per-

per-

and

use

As

the

As

mari-

juana use and sex, age, college year,

size of community of origin, or fathers

education. Marital status and race could

not be adequately examined because 97

percent of the sample were white and 95

percent were single.

8/8/2019 Ginsberg 1978

35/57

As Table 1 shows, involvement in con-

ventional activities such as university

course work, campus organizations, and

employment is not significantly related to

drug use as reported either at the initial or

follow-up interviews. None of the indi-

vidual items used to construct the meas-

ure of involvement in conventional activi-

ties is significantly related to marijuana

use at follow-up,, and further analyses re-

vealed no suppressor effects due to rela-

tionships with other variables. The degree

of involvement in conventional activities,

as measured here, does not contribute to

our understanding of student marijuanause.

Marijuana use is significantly and sub-

stantially higher among students who ad-

mire and identify with marijuana users

(see Table 1). Also, students who have

less commitment to conventional institu-

tions (religious, political, and economic)

are significantly more likely to report

higher frequencies of marijuana use than

are more committed students. Finally,marijuana use is significantly higher

among students reporting higher levels of

psychological distress. Thus the

prior to the initial survey.

Use of marijuana was significantly

higher among students whose fathers had

higher-status occupations as indicated by

25percent of the 274 follow-up

study respondents said they used

marijuana

more times during theprevious

8/8/2019 Ginsberg 1978

36/57

during this period. Forexample,

the (Reiss,

mothers orientation relationships Duncan were (see and occupational were lower ProtestantTable found among 1). between

No status in those significant

religious

whose

index

order correlations shown in Table 1 pro-

vide evidence in support of reference

group theory, commitment theory, and

stress theory.

Our longitudinal data allow an assess-

the responding 479 freshmen used and marijuana

sophomores

or more times in

8/8/2019 Ginsberg 1978

37/57

8/8/2019 Ginsberg 1978

38/57

8/8/2019 Ginsberg 1978

39/57

ment of the relative importance of thesevarious factors with regard to marijuana

use; this analysis is presented in the path

analysis represented in Figure 1.

As might be expected, Figure 1 shows

that the best predictor of marijuana use at

the follow-up survey is marijuana use at

the initial survey. Having a

using reference group is the only other

variable in the model that remains signifi-cantly related to respondents marijuana

use at follow-up. With the other variables

in the

controlled, neither commit-

ment to conventional institutions nor psy-

chological distress is significantly related

to marijuana use at follow-up.

Fathers occupational status has no sig-

nificant direct effect in the path model and

only one indirect effect, due to marijuana

users at Time 1 having fathers with signifi-

cantly higher occupational statuses than

nonusers. marijuana small Religious background indirect effect on students

has a

use through its relationship

with marijuana use at Time 1, i.e., stu-

dents with Protestant mothers report less

28 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

TABLE 1. CORRELATION MATRIX OF STUDY VARIABLES

lVariables 1 2 3 4

Time 2, Follow-up Survey

8. Marijuana-Using Reference

Group (High)

l

8/8/2019 Ginsberg 1978

40/57

11. Respondents Marijuana

Use (High)

Time 1, Initial Survey

3. Involvement in Activities

10. Psychological Distress

(High)

Variables

1. Fathers Occupational Status

(High)

2. Mothers

4. Marijuana-Using Reference

Group (High)

6. Psychological Distress (High)

7. Respondents Marijuana

Use (High)

9. Commitment to Conven-

tional Institutions (High)

Commitment to Conven-

tional Institutions (High)

tant)

correlations of

are significant with

than students

with non-Protestant mothers. In addition,

fewer students with Protestant mothers

claim a marijuana-using reference group at

Time 1 than do students with mothers of

other religious backgrounds. Thus,

mothers Protestant background is also

linked to respondents follow-up mari-

juana use through the reference-group

orientation variable.

Further analyses were done to deline-

8/8/2019 Ginsberg 1978

41/57

ate more carefully the factors leading

to choice of a marijuana-using reference

group and to determine if reference group

also, predicts the other theoretical vari-

ables. As shown in Table 2, reports by a

student that he or she has a

marijuana use at Time

using reference group at follow-up were

regressed on those Time 1 variables in

Figure 1. Reporting a marijuana-using

reference group at time-one is the best

predictor of reporting a marijuana-using

reference group at follow-up. When statis-

tical controls for the other independentvariables are applied, respondents

marijuana use at Time 1 is the only other

1

6 7 8 9

11

8/8/2019 Ginsberg 1978

42/57

8/8/2019 Ginsberg 1978

43/57

8/8/2019 Ginsberg 1978

44/57

variable significantly related to having a

marijuana-using reference group at

follow-up. Neither psychological distressat Time

nor commitment to conventional

institutions at Time

is significantly re-

lated to having a marijuana-using refer-

ence group at follow-up, controlling for

reference-group orientation and marijuana

use at Time 1.

As shown in the regression analysis in

Table 2, commitment to conventional in-

level. Conse-

quently, commitment to conventional in-

stitutions may be spuriously related to

TABLE 2. REGRESSION OF CERTAIN FOLLOW-UP VARIABLES ON TIME 1 VARIABLES

Follow-Up Variables

Time 1 Variables

Commitment to

Marijuana-Using Conventional Psychological

Reference Group

Commitment to Conventional Institutions

(High)

are significant

at p

Marijuana

Use

Respondents Marijuana Use (High)

Psychological Distress (High)

Commitment to conventional institutions as measured at follow-up is a two-item index consisting

of those items on political orientations and religious practices which were included in this variable

measured at time-one.

Distress

FIGURE 1. PATH ANALYSIS OF VARIABLES RELATED TO MARIJUANA USE

Note: Variance unexplained by variables in this model is for the endogenous variables: respondents

marijuana use at follow-up,

8/8/2019 Ginsberg 1978

45/57

Marijuana-Using Reference Group (High)

commitment to conventional activities,

Coefficient is significant at

Standardized regression coefficients; N-274.

Background Factors Initial Survey-Time 1

N =

THEORIES OF MARIJUANA USE

respondents marijuana use at Time 1,

psychological distress,

stitutions at Time

is the only significant

independent predictor of commitment to

conventional institutions at follow-up.

However, there is a tendency for students

reporting a marijuana-using reference

group at Time

to be less committed to

conventional institutions at follow-up, al-though the relationship is small and sig-

nificant at only the p

Coefficients with

marijuana-using reference group,

Distress

16

17

.oo

8/8/2019 Ginsberg 1978

46/57

8/8/2019 Ginsberg 1978

47/57

8/8/2019 Ginsberg 1978

48/57

using reference group at follow-up. Thus,our multiple regression analyses suggest

that these factors are engaged in a recip-

rocal causal relationship of approximately

equal strength.

The relatively equal importance of

marijuana use and reference-group iden-

tification in their reciprocal relationship is

is significantly and positively correlated

with both distress at Timeand identifica-

tion with a marijuana-using reference

group at Time 1, both of which predict

positively to distress at follow-up, thereby

producing a spurious positive correlation

between initial frequency of marijuana use

and psychological distress at follow-up.

Thus our data indicate that, consistent

with the claims of many users, using

up marijuana Regression on the Time

may of actually variables marijuana alleviate (Table use at

2) distress.

shows

the coefficients to differ little from those

of the path model (Figure 1) which in

addition incorporates two background fac-

tors. As Table 2 indicates, identification

with a marijuana-using reference group at

Time 1 predicts marijuana use at follow-up

only marginally better than marijuana use

at Time 1 predicts reporting a

30 JOURNAL OF HEALTH AND SOCIAL BEHAVIOR

marijuana use at follow-up due to the ef-

fect of reference group identification on

both commitment and marijuana use.

Psychological distress measured at

8/8/2019 Ginsberg 1978

49/57

time-one is substantially and significantly

related to psychological distress at the

follow-up, providing further evidence for

the general consistency over time of these

symptom-based 1975). The data further measures indicate (cf. that Srole,

hav-ing a marijuana-using reference group at

Time

significantly predicts psycholog-

for ical distress other factors, at follow-up, individuals i.e., controlling

with a

marijuana-using reference group are more

likely to report distress at follow-up.

However, despite being positively corre-

lated with distress at follow-up, frequency

of marijuana use at Time

predicts nega-

tively to distress at follow-up, frequency

of marijuana use at Time

predicts nega-

tively to distress at follow-up. This result

is marijuana contrary leads to the to argument increased that distress. the use Ac-

of

cording to our data, this surprising result

occurs because using marijuana at Time

table

further confirmed through

analyses. Depending upon how our vari-

ables are dichotomized for this analysis,

Lazarsfelds (1973) Index of Mutual Effect

varies, although it remains relatively low,

ranging from

Thus, having a marijuana-using reference

group is a significant factor in beginning as

well as increasing marijuana use. None ofthe other variables, including background

factors also examined, was significantly

and independently related to marijuana

use at follow-up in this reduced sample.

Various tests for non-linear effects re-

vealed none. The data were also examined

8/8/2019 Ginsberg 1978

50/57

for interaction effects and none were

found.

DISCUSSION.

This analysis clearly demonstrates the

importance of marijuana users whom a

194 (a negative

coefficient indicates marijuana use is the

more important variable). However, the

Index of Mutual Effect has several

weaknesses which may lead to inaccurate

causal inferences (Kessler, 1977). There-

fore, Kesslers indices for generating and

preserving effects were also calculated for

our

tables. These results indicate

that initial levels of each dichotomized

variable are more highly predictive of an

increase in the other variable (generating

effect) than of a decrease (preserving

effect) by a ratio of roughly two to one.

In addition, the results do not consistently

show either variable to be causally domi-

nant, thereby supporting the findings

using the Index of Mutual Effects as well

as our multiple regression analyses.

To explore further whether these mod-els might largely be capturing processes

leading marijuana to use, increased rather than or to decreased

beginning

use, we regressed marijuana use at

follow-up on the Time

variables for

those 127 students reporting no marijuana

use at Time 1. Among this subsample, the

standardized regression coefficients (pre-

dicting marijuana use at follow-up) from

the following Time 1 variables are:

marijuana-using reference group,

commitment to conventional institutions,

and psychological distress,

8/8/2019 Ginsberg 1978

51/57

to

8/8/2019 Ginsberg 1978

52/57

8/8/2019 Ginsberg 1978

53/57

8/8/2019 Ginsberg 1978

54/57

8/8/2019 Ginsberg 1978

55/57

1972; Lavenhar

et al., 1972; Kandel, 1974).

The arguments made by Keniston

it lacks the tested validity and re-

liability of many of the other variables and

thus may not appear as a more important

variable in our analysis. However, com-

mitment as measured here is substantially

correlated with marijuana-using reference

group identification1 and the use of

marijuana. Our study thus replicates past

cross-sectional survey findings. Yet, the

multivariate analysis of our longitudinal

data suggests that commitment does not

significantlyeither choice of a

marijuana-using reference group or fre-

quency of marijuana use over time.

Marijuana use does not appear to be an

anti-social act engaged in to reflect nega-

tive feelings towards the larger society.

Rather, as the importance of

group identification suggests, marijuana

Gergen et al.

use appears to be a positive act designed

to emulate others and probably to help

integrate the marijuana user into his peer

group (cf. Thomas et al., 1975).

The data provide no support for in-

volvement theory. Marijuana use may not

only be deterred by an active life but may

serve as a recreational respite from con-ventional activities. Furthermore, mari-

juana use, unlike some forms of de-

viant behavior, takes relatively little time

and may even be incorporated in the

course of these conventional activities.

Psychological distress as measured in

this study was not found to predict

8/8/2019 Ginsberg 1978

56/57

marijuana use over time independent of

other factors. Proponents of stress theory,

such as

may have prematurely concluded from

cross-sectional data that distress brought

about marijuana use or vice-versa. Our

longitudinal data suggest that, controlling

for other factors, the greaterthe fre-

quency of marijuana use at Time 1, the

fess distress reported at follow-up. That

psychological distress is correlated pos-

itively on a zero-order level with

marijuana use is to some degree explained

by the strong positive correlation between

reference-group identification and the useof marijuana, and the positive effect of

reference-group identification on psycho-

logical distress. Involvement in a

marijuana-using subculture may bring

about changes in living arrangements, re-

lationships with parents, life goals, and

views of self, all of which may be associ-

ated with greater reported psychological

distress.

Some of the modest but statistically sig-

nificant relationships between marijuana

use and socio-demographic factors that

have been previously reported were not

found in this study. We suspect that some

of these relationships, such as the rela-

tionship with sex, have in fact diminished

as marijuana use has spread more perva-

sively through student communities.

The conclusions drawn from this

analysis should be interpreted cautiously.

Although moderate violations of the as-sumptions underlying our analytical

model are not likely to invalidate the re-

sults, this is true for the relative mag-

nitude of effects, not for their absolute

and Louria

8/8/2019 Ginsberg 1978

57/57