Date of Award


Document Type


Degree Name

PhD in Business


Department of Mathematical Sciences: Business Analytics

First Advisor

Dominique Haughton

Second Advisor

M. Lynne Markus

Third Advisor

Heikki Topi

Fourth Advisor

Anne L. Washington


Social network analysis (SNA) is a set of methods used for the examination of the relations found in social structures. While SNA has been used to study business for over 100 years, with early work showing the structure of organizational charts, it has experienced resurgent interest recently with the advances in computing power that allow for much more complex examination of these networks.

This research demonstrates how the use of SNA yields novel insights in three different situations. In study 1 we apply SNA to take a fresh look at U.S. State gubernatorial power. We introduce and implement a weighted network model by which state agency appointments can be examined. Instead of taking a governor-centric approach, as has been the practice, we construct and examine the whole appointment network. Our work shows that continuing with the existing practices will yield misleading results; we propose an alternative and more holistic view of these networks which better illustrates the changing nature of the structure of state government. In study 2 we explore and compare interlocked corporate boards in the U.S. and Europe over a period of 10 years (2001-2010). This longitudinal study examines, through the lens of the interlocked board network, whether the Mizruchi hypothesis, according to which the power of the corporate elite is disintegrating, holds. In study 3, we continue with the theme of interlocked boards but now consider the problem of how to test for statistical significance in network change over time. Our proposed model extends a Bayesian model beyond a pairwise analysis and allows for testing over a multi-year period. We apply and test our model with the interlocked director network in the U.S. over a period of 10 years (2001-2010), but this model is domain independent and can be applied anywhere a network is being examined longitudinally.