Networks Parade

Shahryar Minhas:
  • Associate Professor
  • Department of Political Science
  • Social Science Data Analytics Initiative

Broad strokes

Generally, my research interests falls into two areas:

  • How can we do inference in the presence of interdependent observations
  • Estimating and utilizing an underlying "social space" from an observed network


standard dyadic design

Much of international relations data consists of:

  • a set of units or nodes
  • a set of measurements, yijy_{ij}


the problem

GLM: yijβTXij+eijy_{ij} \sim \beta^{T} X_{ij} + e_{ij}

Networks typically show evidence against independence of {eij:ije_{ij} : i \neq j}

Not accounting for dependence can lead to:

  • biased effects estimation
  • uncalibrated confidence intervals
  • poor predictive performance
  • inaccurate description of network phenomena


evaluation of solutions


Measuring effect of boko haram

with Cassy Dorff & Max Gallop (forthcoming in Journal of Politics)

Measuring Social Space

  • Many political theories often refer to constructs that can not be observed directly
  • As a result, political scientists have come up with latent indicators of everything from:
    • ideological disposition of survey respondents
    • legislators
    • judges

Social space

asymmetric relations from a symmetric network

with Arturas Rozenas & John Ahlquist (Political Analysis [2019])

Social space

strategic determinants of foreign aid

with Cindy Cheng (forthcoming in British Journal of Political Science)

Social space

ccp advancement

with Narisong Huhe & Max Gallop

Social space

predicting subnational intrastate conflict

with Cassy Dorff & Max Gallop (Forthcoming International Studies Quarterly)

multilayer nets

  • Networks that evolve over time or sets of networks that may evolve together over time provide interesting modeling opportunities

Multilayer networks

evolution of cooperation and conflict

with Peter Hoff & Michael Ward (Journal of Peace Research [2016])

Multilayer networks

Measuring state preference

with Max Gallop

Multilayer networks

Measuring Influence

with Peter Hoff & Michael Ward

What's next

  • Mostly, continuing along the lines of what I've discussed
  • Also playing with agent based models a bit to derive some empirically testable insights