Glossary
Social contagion
What is social contagion?
Social contagion describes influence through which exposure to other people affects behavior, ideas, or emotions. The word “contagion” is useful imagery, but similar behavior among friends does not, by itself, show that anything spread.
Influence or similarity?
Friends may influence each other. They may also choose similar friends or share conditions that affect them both. The tendency for similar people to form connections is called homophily.
Christakis and Fowler reported clustering and associations in obesity within a longitudinal social network. Those observations should not be read as proof that obesity spreads like an infection or that a particular friend causes a fixed increase in another person's risk.
Shalizi and Thomas show how observational network data can mix influence with homophily and other causes. Distinguishing them requires strong assumptions or additional causal evidence.
One network, three possible explanations
Imagine that several friends start using the same study app:
- Influence: one friend demonstrates the app, and another starts using it because of that interaction.
- Homophily: students who already liked digital study tools became friends; their similar app choices reflect preferences that preceded the friendship.
- A shared cause: their teacher requires the app, so several friends adopt it independently of one another’s behavior.
All three stories could produce a cluster of app users in a friendship network. Recording when friendships and app use began would help check the sequence, but sequence alone would not settle the cause.
A stronger test could randomly offer an app demonstration to selected students, then compare adoption among their untreated friends with adoption among friends of students not offered it. Researchers would still need to account for overlapping friendship groups and other routes of exposure. The useful comparison is whether changing one person’s exposure changes other people’s behavior.
Why it matters
If a program targets influential people in the hope that change will spread, test whether it does. A connected network is not enough to establish a multiplying effect.