Diffusion of Intervention Effects from Peer Leaders to Their Friends: Effects from the Above the Influence Substance Use Prevention Program

Introduction: Interventions that use peer opinion leaders to deliver program content are often more effective than teacher-led interventions. These improved effects may occur as peer leaders adopt the attitudes and behaviors promoted by the intervention and then influence their friends, thus facilitating the diffusion of intervention effects from peer leaders to the broader population. We use social network analysis (SNA) to examine reach of a school-based peer-led intervention that was developed to integrate with the online Above the Influence social media program. We then test whether proportion of friends who were peer leaders (“peer leader friends”) was associated with changes in non-peer leaders’ substance use intentions over a 5-month intervention period. 

Method: All 8^thgraders at 3 rural middle schools were invited to complete an online survey in January (baseline) and May (posttest). After the baseline survey, 53 peer leaders were trained to carry out a school-wide anti-substance use campaign to decrease perceived social rewards of substance use, increase accurate social norming, and increase efficacy to overcome perceived negative social influences. We focus on 435 students who were not peer leaders (43% Female; 93% White, 4% Latino, 3% Other). On each survey, students named up to 7 friends, which we used to calculate how many steps each student was from a peer leader (1 step = named a peer leader as a friend; 2 steps = named a friend who was friends with a peer leader) as well as the proportion of peer leader friends. Students also reported their substance use intentions (how likely they were to smoke cigarettes, drink alcohol, and smoke marijuana in the next 12 months), which we averaged into a single scale.

Results: At baseline, 43% of non-peer leaders were 1 step and 74.4% were 2 steps from a peer leader. Reach was unrelated to baseline intentions, suggesting that low and high risk students were equally likely to have peer leader friends. Preliminary analyses showed that after 5 months of peer leader programming, students with more peer leader friends reported fewer substance use intentions, even after controlling for baseline intentions and number of intervention activities completed. Future analyses will test differential impact across substances and whether there is an interaction between direct exposure and proportion of peer leader friends.

Conclusions: Future peer-led interventions should try to maximize peer leader reach to the student population to facilitate the diffusion of intervention effects. How peer leaders can tailor their influence to meet specific prevention needs of their friends is another priority for future work. We conclude by discussing the how SNA can be used to study diffusion.