An m-STAR Model of Dynamic, Endogenous Interdependence – a.k.a. Network-Behavior Coevolution – in the Social Sciences

摘要: Spatial interdependence, the interdependence of outcomes across units, is theoretically and substantively ubiquitous central social sciences. The empirical clustering or correlation on some dimension(s), spatial association, also obvious in most contexts. However, may exhibit association for three distinct reasons. First, units be responding similarly to similar exposure exogenous internal/domestic external/foreign stimuli (common exposure); second, units’ responses depend others’ (contagion). A third possibility arises when putative outcome affects variable along which occurs (selection: e.g., homophily). Severe difficulties confront accurate estimation distinction these alternative sources association. After brief review our previous work specification, estimation, testing, interpretation spatiotemporal autoregressive (SAR STAR) models, reflect directly so can address Galton’s Problem distinguishing common from contagion as substantive observed this paper extends those analyses, proposing apply multiparametric (m-STAR) model a simple approach estimating jointly pattern connectivity strength by that pattern, including case where endogenous dependent (i.e., selection). As before, we stress substantivelytheoretically guided structural) specifications support analyses estimated stochastic covariate shocks distinguish possible now three: exposure, contagion, selection. In addition discussing m-STAR compares extant longitudinal-network strategies [this still pending yet], suggests how calculate, interpret, present dynamic, coevolution network structure common-exposure effects emerges such system nonlinear equations. We illustrate interdependence—which parallels models behavior dynamic longitudinal networks literature—with an application attempting disentangle roles economic correlated external internal stimuli, EU membership shaping labor-market policies recent years. * This research supported part NSF grant #0318045. thank Chris Achen, Jim Alt, Kenichi Ariga, Klaus Armingeon, Neal Beck, Jake Bowers, Caporaso, Kerwin Charles, Bryce Corrigan, Tom Cusack, David Darmofal, Jakob de Haan, Scott De Marchi, John Dinardo, Zach Elkins, James Fowler, Freeman, Fabrizio Gilardi, Kristian Gleditsch, Mark Hallerberg, Jackson, Jonathan Katz, Kayser, Achim Kemmerling, Gary King, Hasan Kirmanoglu, Herbert Kitschelt, Kuklinski, Don Lacombe, Lawrence, Tse-Min Lin, Xiaobo Lu, McClurg, Walter Mebane, Covadonga Meseguer, Michael Peress, Thomas Pluemper, Prosperi, Dennis Quinn, Megan Reif, Frances Rosenbluth, Ken Scheve, Phil Schrodt, Chuck Shipan, Beth Simmons, Siroky, D. Stephens, Duane Swank, Wendy Tam-Cho, Vera Troeger, Craig Volden, Ward, Greg Wawro, Erik Wibbels – many others have surely forgotten useful comments and/or other broader project spatialeconometric political science. All remaining errors are ours alone.