When peer pressure changes who people identify with

An intervention can strengthen peer influence and still lower an individual’s eventual action. That is the central possibility explored in Endogenous Identity in a Social Network, a theoretical preprint by Christian Ghiglino and Nicole Tabasso. The mechanism is not simply that people resist pressure: they can change identity, adopting a different behavioral norm and a different reference group.

The paper develops a mathematical model and numerical examples. Its findings establish what can happen within that framework, rather than measuring how often identity changes drive real-world responses to peer interventions.

Belonging changes who matters

Individuals occupy a fixed social network and choose an identity before choosing an action, which could represent educational investment. Each identity carries a status benefit and a prescribed action level. Individuals also benefit from having neighbors who share their identity.

Behavior reflects two conformity pressures: matching the identity’s prescribed norm and matching the average action of same-identity neighbors. Other neighbors remain connected but do not enter the behavioral reference group. Individual ability determines the cost of taking action.

For any fixed identity profile, the model produces a unique equilibrium in actions. Identity choices are more complicated because individuals anticipate the behavior associated with each possible identity. A group’s attractiveness therefore depends not only on its status and local support, but also on how costly its members’ behavior would be to match.

This creates an endogenous adoption threshold. Extra social support can favor an identity, while changes in its members’ behavior can simultaneously make it less appealing.

A new connection can reverse an expected gain

One numerical example uses three individuals. A lower-ability individual is connected to two higher-ability individuals who initially have no connection to each other. All three choose identity A.

Adding a link between the higher-ability individuals raises everyone’s action if identities stay fixed. The lower-ability individual’s action increases from about 1.634 to 1.736.

But allowing identity choice changes the outcome. The added link raises the behavioral pressure associated with A enough that the lower-ability individual switches to B. Their equilibrium action then falls to about 0.833, below its original level. Their own number of connections has not changed; the behavioral consequences of belonging have.

This is an individual reversal, not an aggregate decline in that example: total action rises. Other examples show that assigning an additional person to an identity can prompt existing members to leave, and that rewiring toward greater similarity among connected individuals can trigger identity change rather than entrench existing choices.

Identical abilities do not guarantee identical behavior

The authors also analyze a benchmark in which everyone has the same ability. Here, equilibrium actions are identical within each identity, so pressure to match same-identity peers creates no conformity loss. Differences in local social support can nevertheless sustain different identities. When their prescribed norms differ, actions differ too.

An identity with a lower core value—its payoff excluding direct same-identity connection benefits—can persist if its members have sufficiently stronger support inside their group than outside it. Having many connections is not enough by itself to explain coexistence; their distribution matters.

Starting from everyone choosing the lower-core-value identity, the authors consider a common improvement in the alternative’s attractiveness. Under simultaneous, myopic best-response updating and excluding indifference, individuals with the fewest connections become willing to switch first. Subsequent switches weaken their neighbors’ support for the original identity. Diffusion stops if a sufficiently embedded group can resist the alternative.

More attractive does not always mean higher welfare

In this homogeneous benchmark, universal adoption of the higher-core-value identity maximizes aggregate welfare. A partial transition need not improve it.

Adopters gain from the alternative’s higher core value, but every connection crossing the resulting identity boundary loses its same-identity benefit at both ends. Those losses can outweigh adopters’ gains. Later adopters can even be worse off than before the transition, despite switching being their best response to an already changed environment.

A mechanism, not an empirical diagnosis

The paper connects these results to research on cross-class friendships and deliberately designed student peer groups. A numerical educational example shows how reassignment can raise low-ability students’ actions with identities fixed, yet lower them once those students switch to a lower-norm identity.

The example’s parameters are not estimated from the cited student data, and its actions are not GPA measurements. The empirical studies do not establish the paper’s identity mechanism. Its contribution is to show why observable ability and assigned connections may be insufficient for predicting peer effects: identities, perceived norms and effective reference groups can change alongside social exposure.