Sex differences in social networks
14
groups were significantly larger than those of females (F(1,28)=19.73, p< .001, d = .90 at T1;
F(1,27)=6.10, p< .05, d = .92 at T2). In each school class and at both time points there were
approximately two male groups, one very large and the other small (see Figure 1 Networks 4 and 3).
The number of girls’ groups varied from a single group in one class to nine in another. The smaller
male networks were similar in size to those of females and always consisted of between two and
four members (see Figure 1 Network 3).
To test whether boy’s larger networks are due to them playing team games, we re-analysed
play network data from the 6249 observations of social interaction to determine whether the
exclusion of team game data influenced the mean size of these play networks (see Table 1).
Analyses indicated sex differences in the size of play networks for all games (F(1,6247)=240.32; p<
.001, d = .30) and team games (F(1,1001)=81.64; p< .001, d = .70) but not for play networks when
team game data were excluded (F(1,5197)=0.54; p= .46).
Re-identification of the T1 group level social networks using the play network data for non-
team games (i.e. with team games removed) indicated little change in network size or membership.
Boys’ social networks averaged 6.90 members (SD 2.9) and girls’ networks averaged 3.14 (SD 1.4)
members and an analysis indicated that these social networks were not significantly different in size
from those derived using all of the play network data (F(1,27)=2.13, p= .16).
Social network structure
To examine whether there are sex differences in the internal structuring of social networks
and whether boys spread themselves across a wide range of play partners, we examined social
networks across the three threshold levels of involvement. The data in Table 1 show that across the
three thresholds and at both time points, male social networks were more likely to increase in size
whereas female networks were more likely to increase in number. Focusing first on social network
size, sex differences were evident at cluster and group but not core levels (Clusters T1 -
F(1,25)=6.87, p< .05, d = 1.02; Whole-networks T1 - F(1,28)=19.73, p< .001, d = .90; Clusters T2 -