Sex differences in social networks
12
classified as a 'core' and for 37.5% or more of their scans a 'cluster'. Groups were therefore the most,
and cores the least, inclusive (see Figure 1).
Logical rules were implemented to identify the boundaries between social networks, when a
member was shared between two networks and to prevent the existence of networks based on
sequentially linked individuals (e.g. A connected to B and B connected to C in the absence of a
connection between A and C) becoming incorporated as members of a social network. A core,
cluster or group can have just 2 members. For additional members to be included they must either
be: linked to at least 2 members in the network at that level, or be linked to 1 established member at
that level AND be linked to another at the next level down (i.e. cluster, group or approaching group
level which related to >17% of the combined observations). If a pupil is connected to another
established member of a network but not to any others within the network (in terms of the above
criteria), then this would result in a new network being identified with the established member
shared between the two networks.
The term ‘social network’ is used in the literature to refer to a range of relationships between
people (e.g. friendship networks, collections of individuals that hang around etc.). In the current
study, a social network is the web of inter-connected persons that arise from imposing a threshold
on the aggregated play network data. We refer to networks resulting from imposing a 50% threshold
as cores, 37.5% as clusters and 25% as groups. A network at the group level is considered to be the
‘whole network’ and is equivalent to the notion of a ‘social network’ (‘peer-group’ or ‘clique’) as
currently used by researchers in this field (Cairns et al., 1998; Kindermann, 1993).
To determine the overlap between social networks and best friend and friendship relations,
connections were compared on a dyadic basis. For instance, if two children were connected as part
of a network, their data was also examined to see whether they were connected by a friendship. This
then provided a proportion of the total group level relations that were also friends.
Social network stability was measured in two ways. The first method examined the stability
of each individual’s connections with others in their social network (recall that a connection only