The Sensitivity of Spatial Regression Models to Network Misspecification

Abstract

Spatial econometric models become increasingly popular in various subfields of political science. However, the necessity to specify the underlying network of dependencies, denoted by $\boldsymbol{W}$, prior to estimation is a prevalent source of criticism since the true dependence structure is rarely known and theories mostly provide insufficient guidance. The present study investigates the effects of this network uncertainty which is a special case of model uncertainty that arises from uncertainty about the correct specification of $\boldsymbol{W}$. It advocates Bayesian model averaging (BMA) as a superior approach to this problem, located at the intersection of theory and empirics. Conducting Monte Carlo experiments, I demonstrate that, while the effect estimates are robust toward a misspecification in the functional form of $\boldsymbol{W}$, uncertainty in the neighborhood definition can bias the effect estimates derived from spatial autoregressive models. In contrast to alternative techniques, BMA directly addresses network uncertainty, correctly identifies the true network structure in the set of feasible alternatives, and provides unbiased effect estimates. Two replication studies from different subfields of the discipline illustrate the benefits of this approach for applied research.

STADL Up! The Spatiotemporal Autoregressive Distributed Lag Model for TSCS Data Analysis

“…Likewise, cross-validation and out-of-sample performance are the gold-standard safeguards against overfitting and are similarly extendable to spatiotemporal TSCS contexts. On measurement error, various Bayesian strategies of model averaging (Juhl 2020) and/or combining measurement and estimation models seem most promising to us. These projects head our research agenda going forward.…”

“… 10 Given uncertainty over the relevant ties/network, a Bayesian model-averaging approach to estimating W simultaneously with a model of its effect seems promising (Juhl 2020). …”

A multilevel model for coalition governments: Uncovering dependencies within and across governments due to parties

“…In conclusion, this paper contributes to the literature on coalition governments by dissecting the analytical levels of explanations of coalition outcomes and the link between those levels. The paper also proposes a novel way to model the interdependence structure of political processes, not wholly unlike the spatial autoregressive model (SARM) (Juhl 2020). The difference is that the SARM models how parties affect each other's outcomes while the MMMM models interdependences in the joint effect of parties on outcomes at a higher level.…”

What makes a “regime complex” complex? It depends

“…In this subsection, I treat spatial and social structural dependencies together, as their cross-sectional quality and emphasis on neighborhoods make them analytically similar for our purposes. Moreover, spatial dependencies are often treated as networks to allow more flexible specification of what neighborhood or adjacency means (Juhl, 2019).…”