Identification of Peer Effects in Networked Panel Data

20 Dec 2016

Excited to share that my paper with Daniel Rock and Sinan Aral, “Identification of Peer Effects in Networked Panel Data” was published in the Proceedings of the 2016 International Conference on Information Systems.

This work has a long history going back to early in graduate school for me.  Researchers are always looking for tricks to detect peer effects in observational data, and estimating time-dependent models with longitudinal data has always seemed like a promising approach.  We prove some results about a hybrid spatial-autoregressive model, extending the classic Bramoullé et al. (2009) results to the dynamic setting and estimating the model on some large-scale data.  Our results are somewhat pessimistic, in that it takes fairly strong (basically unrealistic) assumptions to identify a causal effect using panel data models. 

Here’s the abstract:

After product adoption, consumers make decisions about continued use. These choices can be influenced by peer decisions in networks, but identifying causal peer influence effects is challenging. Correlations in peer behavior may be driven by correlated effects, exogenous consumer and peer characteristics, or endogenous peer effects of behavior (Manski 1993). Extending the work of Bramoullé et al. (2009), we apply proofs of peer effect identification in networks under a set of exogeneity assumptions for the panel data case. With engagement data for Yahoo Go, a mobile application, we use the network topology of application users in an instrumental variables setup to estimate usage peer effects, comparing a variety of regression models. We find this type of analysis may be useful for ruling out endogenous peer effects as a driver of behavior. Omitted variables and violation of exogeneity assumptions can bias regression coefficients toward finding statistically significant peer effects.

AIS members can read the draft now, but we’ll be sharing a full draft with expanded findings soon.