Thursday, January 19, 2012

Latest and Greatest of Metrics with Networks

A well informed friend told me about the latest MIT grad development students working the metrics behind networks:

Which we all know is majorly hopeless because of the reflection problem, and difficulty creating exogenous shocks to social networks (except in my dissertation :).

There's some gnarly modeling of individuals correlation to other individuals in the var-cov matrix (this is a parametric estimation), as well as a relaxation of assumptions, like independence across individuals' networks in a sample.

No comments:

Post a Comment