This study found:
- The Strengthening Families Program for Youth 10-14 (SFP10-14) reduced substance use among the friends of teens who participated in the intervention, as well as the participants themselves.
- The friends' substance use reductions were mediated by altered attitudes toward substance use and reductions in unsupervised socializing with peers.
In SFP10-14, families with children ages 10 to 14 meet with intervention facilitators once a week for 7 weeks to discuss substance use, parenting practices, communication skills, responses to peer pressure, and other topics. Previous studies have demonstrated that the program reduces participating children's substance use and improves participating parents' parenting practices. The new study evaluated the program's effects on the participating teens' nonparticipating friends.
Dr. Kelly Rulison of the University of North Carolina at Greensboro and colleagues at Pennsylvania State University analyzed data collected from more than 5,400 students who attended sixth grade in 13 rural Pennsylvania and Iowa communities. None of the students participated in SFP10-14, even though the intervention was offered to all sixth graders in their schools. Each year for 3 years, the researchers elicited from each student the names of up to 7 peers in the same grade who were "close" friends. They also collected information on each student’s exposure to friends who participated in SFP10-14, to friends’ positive or negative attitudes about substance use, friends’ smoking or drinking to inebriation, and other variables.
- Text Description of Graphic
The figure shows four line graphs illustrating the relationship between cigarette use by adolescents who had not participated in the SFP10-14 intervention and the number of their friends who did participate in the SFP10-14 intervention. The vertical (y)-axis shows the proportion of nonparticipating adolescents who reported using cigarettes in the past month. The horizontal (x)-axis shows the assessment timepoints: before the SFP10-14 intervention; immediately after the intervention; and at 1 year, 2 years, and 3 years after the intervention. The different graphs show the cigarette use of nonparticipants with zero participating friends (yellow line), 1 participating friend (purple line), 2 participating friends (red line), or 3 participating friends (blue line). Before the SFP10-14 intervention, about 3 percent of nonparticipants in all four groups reported using cigarettes. Immediately after the intervention, the proportion of nonparticipants with zero, 1, or 2 participating friends who reported cigarette use remained more or less the same; only those with 3 participating friends reported an increase in cigarette use to about 7 percent. At the three follow-up assessments, however, the proportion of nonparticipants who reported cigarette use increased steadily in all groups, but was greatest among those with zero participating friends (about 7 percent, 13 percent, and 20 percent at 1, 2, and 3 years, respectively) and slightly lower among those with 1 participating friend (about 6 percent, 10 percent, and 14 percent, respectively) or 2 participating friends (about 6 percent, 8 percent, and 14 percent, respectively). Nonparticipants with three participating friends had the lowest proportion of cigarette users at all three follow-up assessments, namely about 2 percent at 1 year, 5 percent at 2 years, and 8 percent at 3 years.
The researchers' analysis revealed that the benefits of SFP10-14 spread from participants to their friends. Thus, the more participant friends a nonparticipant had, the less likely he or she was to engage in substance use in the years following the intervention. At the 3-year follow-up, nonparticipants who had three or more participant friends were roughly 2/3 as likely to report that they had been drunk in the past month, and roughly 1/3 as likely to have smoked a cigarette in the past month, compared with those who had no participant friends (see Figure).
Two mediating factors accounted for most of the indirect benefit experienced by the SFP10-14 nonparticipants. Most influential was the amount of time they spent "hanging out" with friends without adult supervision. Dr. Rulison says, "Multiple mechanisms for this result are possible, but it's most likely that SFP10-14 changed participating parents' supervision practices. Parents who have participated in the intervention tend to supervise their adolescents closely. Nonparticipating teens who spend time with friends who participate receive indirect supervision from their friends' parents, regardless of how much their own parents supervise them."
SFP10-14 nonparticipants' substance use also was influenced by their participant friends’ attitudes toward smoking and drinking alcohol. Although this effect was small compared to that of unsupervised socializing, it implies that encouraging participants to advocate negative attitudes about substance use to their friends could help reduce community-wide teen substance use.
Additional findings from the study underscore the strong influence that peer behavior can have among teens and the potential for interventions such as SFP10-14, which reduce problem behaviors, to benefit teens who do not directly experience them. The researchers calculated that a unit increase in smoking prevalence among a teen’s friends was associated with a 14-fold increase in his or her odds of smoking, and an increase in the friends' prevalence of drunkenness was associated with a near quintupling of his or her odds of getting drunk. However, the researchers acknowledge that selection processes also play a role in shaping teen behavior—that is, that teens who drink alcohol or smoke gravitate to friends who do the same.
Dr. Rulison notes that all the school districts in the study were majority-white with stable student populations, and the findings may not apply to other types of communities. She comments, "Diffusion results from the stability of the community and changing community norms, not community demographics. Whether diffusion occurs in more transient communities depends on the specifics of the intervention." For example, she says, because the benefits of SFP10-14 spread partly by altering the behavior of participating parents, "diffusion is less likely if participating parents move away.”
However, the researchers also believe that diffusion may occur via the cumulative, normative effect of students' beliefs. "Changing individual attitudes could lead to a sustained school- or community-wide change in norms, even if many of the original program participants move away," Dr. Rulison says.
The researchers say that identifying the specific mechanisms and processes that support diffusion of a programs' benefits can enable researchers to improve in program design and implementation. Accordingly, they recommend that program developers and evaluators measure their programs’ impact, if any, on nonparticipants, such as those who join the community after the intervention, siblings of participants, and nonparticipants who are not in the same class or grade in which the program is implemented.
Dr. Rulison and colleagues advise intervention designers to leverage diffusion effects to maximize their programs' impact. "Intervention developers should target factors, such as peer attitudes and unstructured socializing, that might facilitate diffusion," Dr. Rulison says. "Some programs already do so by specifically training student leaders to spread intervention messages."
This study was supported by NIH grants DA018225, DA013709, HD041025, AA14702, and the WT Grant Foundation.
Rulison, K.L.; Feinberg, M.; Gest, S.D.; and Osgood, D.W. Diffusion of intervention effects: The impact of a family-based substance use prevention program on friends of participants. Journal of Adolescent Health 57(4):433-440, 2015. Abstract