Birds of a Feather — Does Similarity Really Drive Friendship?

What Is Homophily?

Homophily describes the tendency to befriend similar people. McPherson, Smith-Lovin, and Cook (2001) showed in the most influential review on the topic: similarity is particularly strong in demographic characteristics (age, gender, education) and somewhat weaker in values and interests.

Similarity and opportunity structures reinforce each other: future friends meet at university because similar education and interests led them down the same path. Physical proximity is therefore not coincidental but partly a consequence of pre-existing similarity.

Perceived vs. Actual Similarity

The key finding: not actual similarity but perceived similaritypredicts whether two people become friends. Van Zalk and Denissen (2015) showed in a longitudinal study: perceived — but not actual — similarity in personality traits predicted the later formation of friendships.

This means it matters less whether two people are objectively similar than whether they feel similar. Liking someone may lead to retrospectively overestimating similarity. The direction of causality is not always clearly separable.

Which Similarities Matter?

Similarity in activities and interests matters more for friendship formation than similarity in attitudes (Werner & Parmelee, 1979). The reason: shared activities have direct implications for joint interaction.

Selfhout et al. (2010) used network analysis to show that overall similarity is less important than domain-specific similarity in extraversion (matching interaction style), agreeableness (matching cooperativeness), and openness (shared values and interests).

Among children and adolescents, similarity in prosocial and antisocial behavioris a particularly strong predictor — aggressive children actively seek out other aggressive children as friends as early as preschool age (Haselager et al., 1998).

The Cost of Sameness

Homophily is not all upside. Gompers, Mukharlyamov, and Xuan (2016) showed: venture capital investors with similar backgrounds collaborate more often but make significantly worse decisions— an effect attributed to groupthink and a lack of diverse perspectives.

From a sociological perspective, there is another downside: friendships prove to be homogeneous across all major socio-structural dimensions — class, milieu, age, gender, and status (Alleweldt, 2016). Friendships thus serve a “gatekeeping function”: social strata tend to stay among themselves. The supposedly free choice is structurally constrained.

Genetic Similarity

Similarity goes deeper than education and values. Christakis and Fowler (2014) showed: friends are genetically more similar than strangers — at the level of fourth cousins. Homophily is especially concentrated in olfactory receptor genes, while immune-system genes show heterophily.

This biological dimension of friend selection adds an unconscious factor to the homophily concept. We choose friends not only based on conscious preferences and social opportunities but partly also based on genetic similarity — without knowing it.

Research Limitations

Whether similarity causes friendship or friendship produces similarity (through socialization effects) cannot be resolved with cross-sectional data. Moreover, recent research shows that similarity across multiple dimensions simultaneously has diminishing returns— “the more similar, the more likely” is too simplistic.

Frequently Asked Questions

Is it true that birds of a feather flock together?

Yes — homophily is one of the most robust findings in friendship research. Friends are more similar than random pairs — in age, gender, education, and values (McPherson et al., 2001). However, perceived similarity matters more than actual similarity.

Which similarities matter most for friendship?

Similarity in activities and interests weighs more heavily than similarity in attitudes, because it has direct implications for shared interaction (Werner & Parmelee, 1979). Among adolescents, similarity in prosocial and antisocial behavior is a particularly strong predictor.

Can similarity also be harmful?

Yes. Gompers et al. (2016) showed that venture capital investors with similar backgrounds collaborate more often but make significantly worse decisions — due to groupthink and a lack of diverse perspectives.

Do we unconsciously choose similar friends?

Partly. Genome-wide analyses show that friends are genetically more similar than strangers — especially in olfactory receptor genes (Christakis & Fowler, 2014). Social structures also channel us into contexts where we encounter similar people — without conscious choice.