PREVENTIVE

MEDICINE

19, 305-313 (1990)

Drug Use Prevention Programs, Evaluation of Three Seventh-Grade

Gender, and Ethnicity: Project SMART Cohorts’

JOHN W. GRAHAM, PH.D.,**’ C. ANDERSON JOHNSON, PH.D.,* WILLIAM B. HANSEN, PH.D.,*,~ BRIAN R. FLAY, D.PHIL.,*,~ AND MIMI GEE M.S., M.D.? *Institute for Health Promotion and Disease Prevention Research and Department of Preventive Medicine, University of Southern California, Pasadena, California 91101; and tSchoo1 of Medicine, University of Southern California, 1975 Zonal Avenue, Los Angeles, California 90033

One-year follow-up data from three seventh-grade cohorts of Project SMART were examined to assess the effects of two social psychology-based programs within each of six subgroups: males, females, Asians, blacks, Hispanics, and whites. The three cohorts (total N = 5,070) were those receiving curriculum or serving as controls as seventh-graders in 1982-1983, 1983-1984, and 1984-1985school years. The outcome measures used were composite indices based on lifetime and recent use items for cigarettes, alcohol, and marijuana. The major analysis was ANCOVA on classroom means for the composite indices at posttest, using pretest classroom means for the indices as a covariate. The results showed clear prevention effects for females but not for males. Overall prevention effects were strongest for cigarette smoking, but were also evident for alcohol. Significant sex by program interactions, showing differential program effects for males and females, were found for cigarettes and marijuana, but not for alcohol. There was a nonsignificant trend suggesting that the programs were most effective for Asians and least effective for whites. 0 1990 Academic Press, Inc.

INTRODUCTION Hansen, Johnson, Flay, Graham, and Sobel(1) showed that a program designed to give seventh-grade students social resistance skills was effective in reducing the onset of cigarette, alcohol, and marijuana use. However, a program designed to give seventh-graders personal decision and stress reduction skills (referred to as an “affect management” program) was not effective. Johnson ei al. (unpublished manuscript) found that a social skills program delivered to sixth-graders was effective in reducing the onset of alcohol and cigarette use. A multiple-year (sixth, seventh, and eighth grade) social skills program and a multiple-year affect management program containing elements of social resistance skills training were also efficacious. Both reports showed clear evidence for the efficacy of social skills programs i Supported by a grant from the National Institute on Drug Abuse, Grant I-R18-DA030496. * To whom requests for reprints should be addressed at USC-IPR, 35 North Lake Ave., Suite 200, Pasadena, CA 91101. 3 Present address: Bowman Gray School of Medicine, Wake Forest University, Department of Health Sciences, 300 South Hawthorne Road, Winston-Salem, NC 27103. 4 Present address: Prevention Research Center, School of Public Health, The University of Illinois at Chicago, 2121 West Taylor Street, Box 6998, Chicago IL 60680. 305 0091-7435190$3.00 Copyright Q 1990 by Academic Press, Inc. All rights of reproduction in any form reserved.

306

GRAHAM ET AL.

aimed at young adolescents in general. It is also important to know for which subgroups of adolescents the programs are most and least efficacious. In the present article, efficacy of the social skills and affect management curricula are examined for males and females, and for each of the four major ethnic groups in the study, Asian, black, Hispanic, and white. METHODS Subjects Subjects were 5,070 seventh-graders participating in Project SMART (SelfManagement and Resistance Training) (l-3). Data are presented for three cohorts-students who received the program (or served as controls) as seventhgraders in the 1982-1983, 1983-1984, or 1984-1985 school years, and the approximately 70% who were present for measurement 1 year later as eighth-graders. Approximately 6% were Asian, 20% were black, 31% were Hispanic, and 43% were white. Research Design and Assignment to Conditions The research design for the three cohorts is presented in Table 1. Sixteen junior high schools were used in all. Schools were randomly selected from a total population of 63 junior schools. Assignment for all cohorts was made prior to the 1982 implementation. The randomizing procedure used, which is described in more detail elsewhere (3), was designed to maximize pretest comparability of groups. Treatment Conditions The details of the treatment conditions have been described elsewhere (1). Briefly, for all three cohorts, the social skills program (SOCIAL) was essentially TABLE 1 EVALUATION DESIGN

Year Cohort

School No.

1982-1983

1982

1, 2 394 5,677, 8 5, 6 7, 8 9, 10, 11, 12 9, 10 11, 12 13, 14, 15, 16

0 A, 0 s, 0 G

1983 1984

1983-1984

1984-1985

1985-1986

0, OS

OfI 0 A, 0 s7 0 G

Note. Subscripts refer to grade of measurement implementation. A refers to Affect program; S refers to social program; C refers to control; 0 denotes measurement.

DRUG

USE PREVENTION

PROGRAMS

307

the same: a 1Zsession program giving students social skills for resisting drug’ offers. These sessions included teaching students about the various sources of social pressure to use drugs, techniques for resisting them, and role-play opportunities for practicing the resistance techniques. For the 1982 cohort, the affect managementprogram (AFFECT) was a 1Zsession program which contained no social skills sessions.These lessonsfocused instead on personal decision-making, values clarification, and stress management techniques. However, for the 1983 and 1984cohorts, the 1Zsession AFFECT program also contained some sessions in basic resistance skills (as described above). For all three cohorts, students in the CONTROL conditions were measured, but received no special drug prevention curriculum other than that already provided by the schools. Measures

Measures were taken via questionnaire in a separate sessionjust prior to the first program lesson. Before completing the questionnaire, students learned that a sample (saliva in 1982-1983and 1983-1984;breath in 1984-1985and 1985-1986) would be requested, which would later be analyzed for saliva thiocyanate (carbon monoxide in 1984-1985 and 1985-1986). Students learned that analysis of the samplewould allow the researchersto know whether they had been smoking. The importance of being honest was stressed, and confidentiality was assured. The outcome measures used in this article were composite indices of lifetime and recent use of cigarettes, alcohol, and marijuana. The cigarette smoking index was based on three items: Number of cigarettes in lifetime’j; number of cigarettes in the past 30 days; and smoking status (e.g., never smoker, previous smoker, etc.). The alcohol use index was based on four items: Number of alcoholic drinks in lifetime,’ number of drinks in the past 30 days; number of drinks in the past 7 days, and number of days in the previous 30 days the person drank alcohol. The marijuana use index was based on number of times marijuana had been used in lifetime, in the past 30 days, and in the past 7 days. An overall “drug use” index was formed by standardizing and averaging the three substance-specific indices [e.g., see (4)]. Program schools were usually pretested earlier in the semester than control schools. Thus, raw pretest scores might be higher in control schools than in treatment schools simply because they were measured later. In order to take this difference into consideration, the major analyses were performed using as a co’ In the present paper, “drug use” refers to alcohol, cigarette, and marijuana use. 6 For all drug use items, subjects were fust asked (e.g., for cigarettes), “Have you ever had even a puff of a cigarette?” Those who responded “yes” to these lead items went on to complete the remaining questions. Subjects who responded “no” to the lead items were assumed to have no use on the remaining items. The lifetime use item for cigarettes was: “How many cigarettes have you smoked in your whole life?” (a) just puffs; (b) ah or part of one; (c) 2 to 4; (d) 5 to 10; (e) 11 to 20; (f) 1 to 5 packs; (g) more than 5 packs. Response categories were the same for the 30-day use item, except that the highest category (f) was “more than 20.” ’ Response categories were: (a) sips; (b) all or part of one; (c) 2 to 4; (d) 5 to 10; (e) 11 to 20; (f) more than 20.

308

GRAHAM

ET AL.

variate the residual of the pretest index regressed on month of measure. Month of measure was equivalent across conditions at post-test. Measures of Gender and Ethnicity

The students indicated whether they were boys or girls. Data collectors categorized each student’s ethnic identity based on appearance, language spoken, and name. The categories were Asian, black, Hispanic, white, Hispanic surname but looks white, and other. Independent assessments of ethnicity were made at each wave of measurement. Analysis

Strategy

The major analysis was analysis of covariance (ANCOVA) using subjects’ posttest score on the drug use index as dependent variable, and pretest score (residual) as covariate. Only analyses using pretest classroom as the unit of analysis are reported (N = 327). Classroom averages were obtained for each of the eight subgroups (Asian male, Asian female, black male, black female, Hispanic male, Hispanic female, white male, and white female). However, almost none of the classrooms had members of all eight subgroups (the average was 4.7 subgroups per class). Thus, there were only 1,520 classroom-subgroups available for analysis. Because some of the classroom-subgroups were small (e.g., N = l), and there were some missing data, sample sizes varied slightly from analysis to analysis. We began by collapsing the two kinds of program for an analyses contrasting (any) program and control groups and then compared one program with the other. Also, we began by examining the results for the overall drug use index and then went on to examine the results separately for smoking, alcohol, and marijuana. Finally, we first presented a single 2 (program vs control) X 2 (male vs female) x 4 (Asian, black, Hispanic, white) ANOVA and then explored program effects among subgroups. RESULTS Reliability

and Validity

Internal consistency for the drug use indices was excellent in the present sample with coefficient OL(5) ranging from 0.86 for the alcohol index at pretest to 0.91 for the marijuana index at post-test. Overall, 91% of the independent ethnicity judgments were consistent from year to year, and were Asian, black, Hispanic, or white. Only these cases were used in the analyses reported. Omnibus Analyses

The first set of analyses examined the composite drug use index using a 2 (program/control) x 2 (sex) x 4 (ethnicity) ANOVA. As predicted, date of measure was a significant predictor of the drug use index at pretest, F(1,1518) = 16.66, P < 0.0001. The drug use index was regressed on date of pretest measure and the residual was used as the pretest for later analyses. Table 2 presents the pretest and post-test means and standard deviations corresponding to main effects and two-way interactions involving the program

DRUG

USE PREVENTION

309

PROGRAMS

TABLE 2 MEANS AND STANDARD DEVIATIONS FOR STANDARDIZED DRUG-USEINDEX

Black

Asian Control

Program

Mean

(SD)

Mean

Pre Post 8

0.20

(0.49) (0.63)

0.40 0.52

Pre

0.24 0.32

Post

0.44

09

Total

he Post

Program Total

Pre Post 00 Pre Post

ET Post WI

0.50 0.67

(0.56)

(0.62) (0.39)

0.47 0.59

(185) (0.38) (0.49)

0.43 0.50

(0.57) (0.45)

(0.46) (0.49)

0.49 0.63

(0.50) (0.52)

(375)

(489)

Male

Total

(SD)

Mean

(SD)

Mean

(SD)

0.42 0.70

(0.50) (0.61)

0.54

(0.56) (0.53)

0.48 0.68

(0.54) (0.58)

(0.54) (0.54)

0.50 0.63

0.43 0.59

(374)

(0.49) (0.51)

0.56 0.67

(0.50) (0.56)

0.55 0.67

(374) 0.43 0.65

(SD)

(0.51) (0.56)

0.48

0.68

(0.54) (0.58)

0.50 0.63

(0.52) (0.52)

(0.52) (0.58)

(235) 0.65 0.91

6’52)

(768)

(0.52) (0.57)

5’52)

(0.52) (0.52)

(768)

(394)

(752)

(SD)

(243) 0.67 0.92

Total Mean

(478)

Mean

0.66

0.62 0.91

(254)

(178)

WE)

Mean

(2351

Female

09

Control

0.45 0.48

(94) 0.22 0.38

(0.49)

(SD) (0.55)

WV

(8.9 (0.26) (0.31)

(SD) (0.51)

White

Hispanic Mean

(0.55) (0.53)

VW

effect.’ Significant pretest differences were found for sex and for ethnicity. Mean drug use among females was less than that among males at pretest. The order of mean drug use at pretest for the four ethnic groups was: Asians (lowest), blacks, Hispanics, and whites (highest). Table 3 presents the P values from the 2 x 2 x 4 ANCOVA using the overall drug use index (post-test) as the dependent variable and the pretest drug use index (residual) as the covariate. Two effects involving the program were significant. The overall main effect for program vs control was significant; program classrooms had lower means for the drug use index than did control classrooms. Also, the sex by program interaction was significant; the difference between program and control was greater for females than for males. As shown in Table 3, the difference was significant for females but not for males. Overall, females reported substantially less drug use than males at pretest, but reported only slightly less use than males at post-test. The sex by ethnicity interaction was significant because the pattern just described held for all ethnic groups but Hispanics. For Hispanic students, the relative drug use between males and females was only slightly changed from pretest to post-test. The significant main effect for ethnicity had a similar interpretation at pretest and post-test; the order of reported drug use was Asians (lowest), blacks, Hispanics, and whites (highest). s The drug use index scores appearing in Table 2 have been arbitrarily resealed to improve interpretability. A constant was added to all pretest and post-test scores so that all subgroup means were positive. Further, a constant was added to all post-test scores so that the post-test value was greater than the pretest value for each subgroup. This resealing had no effect on the analyses performed.

310

GRAHAM ET AL. TABLE 3 SUMMARYOF ANOVA TABLE P VALUES

Pretest effect Ethnicity Sex Program vs control Ethnicity x program Sex X program Ethnicity x sex Ethnicity x sex x prog Post-test effect Ethnicity Sex Program/control Ethnicity x program Sex X program Ethnicity x sex Ethnicity x sex x program Simple effects program/control By ethnic group Asian Black Hispanic White By sex Females Males

Smoking index

Alcohol index

Drug use prevention programs, gender, and ethnicity: evaluation of three seventh-grade Project SMART cohorts.

One-year follow-up data from three seventh-grade cohorts of Project SMART were examined to assess the effects of two social psychology-based programs ...
643KB Sizes 0 Downloads 0 Views