Triad census analysis.
The triad census analysis of the triad distribution showed that the
observed network has a significant excess of transitive triads (N=76)
followed by a significant deficit of cyclical triads (N=8). Triad types
that are positive (i.e. non-overlapping at 0) occurred in excess in the
observed network, while triad types that are negative showed a deficit
in the observed network as compared to the random null network (Fig 2).
The remaining five triads in the network did not show any significant
differences in the mean triad percentage rates between the observed and
expected network. While the data showed a clear excess of transitive
triangles (34.55 %) and deficit for cyclical triangles (3.6%), the
distribution for pass-along triads shows a less typical pattern with the
95% confidence intervals crossing the zero line but the mean percentage
still showing a deficit.