Friendship Reciprocity and Behavioral Change


When we analyzed self-reported relationship surveys from several experiments around the world (from human subjects, not hobbits!), we found that while most people assume friendships to be two-way, only about half of friendships are indeed reciprocal. In itself this may seem like an interesting but minor finding, but this large proportion of asymmetric friendships translates to a major effect on the ability of individuals to persuade others to cooperate or change their behavior.

For example, when we examined the properties of friendship networks and how the directionality of ties can impact the level of influence that individuals exert on one another (based on analysis of a fitness and physical activity intervention where information about physical activity was collected passively by smartphones), we found that the program was more effective when a unilateral friendship tie existed from the buddy (the person applying peer pressure) to the subject (the person receiving the pressure) than when the friendship tie was from the subject to the buddy. In this example, reciprocal friendships are best, but having a buddy who thinks of the subject as a friend is the next best relationship. We attribute the difference to our peer-to-peer incentive mechanism—as buddies were rewarded based on the progress of the subject, there are likely to be differences in communication when the buddy believes the subject to be their friend versus when they do not.

The findings of this work have significant consequences for designing interventions that seek to harness social influence:

Intervention designers, whether with fitness programs, smoking cessation programs, or any other attempt to change a subject’s behavior, can't rely on how the subject perceives the relationship with the buddy to create effectiveness.

Also, we shouldn’t assume people with a high number of social ties are “influencers.” Such people are no better and often are worse than average at exerting social influence. Our results suggest that this is because many of those ties either are not reciprocal or go in the wrong direction, and therefore won’t lead to effective persuasion.

We demonstrate that an assumption common in previous studies of social influence, namely that friendships are created equal or reciprocal by default, is erroneous, which may have significantly biased the research results.

We hope that by understanding the factors and network properties that impact the level of social influence individuals exert on one another, we can be more effective at promoting behavioral change, disseminating new ideas, and even promoting products.