When Do Degree Centrality and Tie Strength Centrality have Divergent Validity

Scholars often debate whether degree centrality–the total number of ties that connect to an actor–or tie strength centrality–the total strength of those ties–is the more appropriate measure of how connected actors are to others in a larger network. This debate may be rendered moot by how researchers collect information about the strength of a network’s ties. A meta-analysis of 125 networks comprising 151,705 actors suggests that degree centrality and tie strength centrality are log-log correlated at ¯r = .94 when researchers assess tie strength by counting the number of times that pairs of actors attend the same events (untransformed ¯r = .95); the measures are log-log correlated at ¯r = .90 when researchers assess tie strength with rating scales (untransformed ¯r = .92); and (c) the measures are log-log correlated at ¯r = .85 when researchers assess tie strength by counting direct connections such as face-to- face interactions or email correspondence (untransformed ¯r = .81). Finding average correlations...