1. The use of over-aggregated racial categories in previous research may heave masked ethnic in-group preferance
2. A failure to take into account opportunity structures and differences in the sociality of the members of these groups led to previous studies over-aggerating the homophily rates of larger groups and those of groups comprised of more sociable individuals with larger networks
3. Previous research into homophily used statistical techniques that ignored triadic closure effects, thus leading to an overestimation of the tendency towards homophily
4. They found other characteristics than basic demographic attribute data

Social influence
Social networks may influence certain behavior via process of information transmission through opinions, attitudes and normative values of the group.


Social selection/homphily
People that are similar are more likely to want to connect with each other. (same sense of music, dressing, opinion, political orientation, religious identity etc.)


Contextual factors
Structure and compositions of a person’s social network. (shared contextual factors that might influence the behaviour  seeing certain adds that might influence the behaviour instead of social influence or selection). Such as smoking and kids, they might both smoke because they have seem the same commercial

Simple contagion: A single “infected” source is sufficient to transmit X
- If one adopter among friends is sufficient for adoption  simple contagion
Complex contagion: multiple sources of “infection” required before being convinced to use X
- If multiple adopters (or a % of adopters) among friends matters  Complex contagion