of a Family-Based Substance Abuse Prevention Program
in Rural Communities
Published in The Journal of Primary Prevention, Vol 18, No. 3, 1998
If a program is not faithfully implemented by staff, then it has little chance of producing statistically significant outcome findings that can be replicated. The evaluation staff made frequent observations of parent and student group sessions. Overall, implementation was very good. Group leaders followed the curriculum and handled discussion questions and role plays well. The most frequent problem was going past the allotted time because activities took longer than anticipated.
A quasi-experimental research design was used (Cook & Campbell, 1979). This was a voluntary program, thus it was particularly important to determine the comparability of parents and youth who chose to participate in the program and those who did not. One-way analyses of variance (Program Graduate versus Nonparticipant) were conducted for all the baseline measures for students and parents who completed the baseline and one-year follow-up survey.
Students'findings. As can be seen in Table 1, students who graduated from the FIA program had at baseline significantly less appropriate attitudes towards adolescent tobacco use, lower scores on family cohesion, lower school attachment and higher rates of talking to counselors than did members of the comparison group. Student graduates also had significantly more school absences and were significantly older than members of the comparison group at baseline. FIA graduates had marginally less appropriate attitudes towards adolescent use of alcohol and lower peer attachment. There were no significant baseline differences for reported "age O.K. to drink alcohol," school activities or grade point average.
Parents' findings. For parents, there were three baseline differences between graduates and members of the comparison group. Table 2 shows that parents who graduated from the FIA program had significantly lower scores at baseline on cohesion and had significantly higher rates of talking to counselors. Also, parent graduates had significantly fewer years of education than did comparison group parents. There were no significant baseline differences in appropriate alcohol attitudes, appropriate tobacco attitudes, reported "age O.K. to drink alcohol," family activities, attachment to their child's school, school activities, family income or number of children in the home.
Because of the baseline differences, all comparisons of the one-year follow-up scores for program graduates and comparison group members controlled for initial differences between groups on the baseline survey. Baseline scores were treated as a covariate so that change from baseline to follow-up was compared for the two groups, and therefore only adjusted means are presented below. Also, demographic information from parents (education, number of children and income) and students (age, absenteeism and grade point average) were treated as covariates. Inclusion of the demographic information as covariates did not change any of the program findings, thus they were excluded from the analyses presented below.
StudentsÕ results. A series of one-way analyses of covariance (ANCOVA) was conducted to examine program effects. There were four significant main effects. A series of 2 (Male, Female) x 2 (Program Graduate, Nonparticipant) ANCOVAS was also conducted to detemiine whether the program effects were comparable for boys and girls. There were approximately an equal number of girl and boy graduates (N = 22 and N = 21, respectively) and girl and boy comparison group members (N = 192 and N = 171, respectively).
Three significant program effects were moderated by gender (appropriate alcohol attitudes, school and peer attachment), therefore, only the interactions are reported below. The main effect of talking to counselors was not moderated by gender. Controlling for baseline scores, student gmduates (M = .40) were more likely than nonparticipants (M = .25) to talk to a counselor at the one-year follow-up, F(1,399) = 8.28, p < .004.
Significant interactions between program participation and gender were found for four student measures. Controlling for baseline scores, at the one-year follow-up boy program graduates scored significantly higher than did boy nonparticipants on: appropriate attitudes towards alcohol, age reported that it is "O.K. to drink alcohol," school attachment and peer attachment (see Figures I & 2). These four program effects were not significant for girls.
Parents' results. As with the students, a series of one-way ANCOVAS was conducted for all outcome measures with parents. When controlling for baseline scores, parent graduates at the one-year follow-up reported more involvement in school activities than did nonparticipants, F(1,201) = 9.93, p < .002; M = .65 and M = .54, respectively. Parent graduates also reported more involvement in family counseling (M = .52) than did nonparticipants (M = .32), F(1,201) = 10.96, p < .001 controlling for baseline scores. No other significant effects were found for parents. Most parent program participants were women (79% female; 21% male), thus gender of participant effects could not be examined for parents.
Students' results. To examine the short-term effects of program participation, paired t tests were conducted using the pretest, posttest and 10-week follow-up data which was collected only for program participants. For students, effects were found on two of the nine measures. Student graduates had significantly higher peer attachment scores at the six-week posttest 9 (M = .76) than at the pretest (M = .70), t(43) 2.16, p < .04. This effect was not significant at the 10-week follow-up.
Students' attitudes about adolescent alcohol use became significantly less socially appropriate from the pretest (M = 3.03) to the posttest (M = 2.74), F(43) = 3.1 1, p < .003. When boys' and girls' data were analyzed separately, this effect was found to be significant only for girls (M = 3.05 pretest; M = 2.59 posttest), t(2 1) = 4.3 1, p < .001 This is presumed to be a maturation effect, rather than a program effect, because girl program participants did not score significantly worse over time than did the comparison group participants (see Figure 1).