A preliminary social network analysis of MPACT

The dissertation author and committee relationships in the MPACT dataset were modeled as social networks. Exploratory network analysis was performed primarily on the network composed of dissertation advisors and advisees to investigate the utility of structural prestige for measuring mentoring impact. Three prestige measures were investigated: 1) outdegree, 2) output domain, and 3) proximity prestige. All three measures were highly correlated when computed across the entire dataset, but correlation was lower for those scoring in the top 25 of any of the three. Correlation with citation ranking was low for this subset.