20 May 2017

Event Sequencing


– “Event Sequences in Disputed Issues” (with Emily U. Schilling) – in progress
– “Order? Order!: A Generalized Sequence Model for Sequence-Related Hypotheses” (with Emily U. Schilling) – in progress


Emily U. Schilling and I extend the idea of stages-as-context by considering stage sequences as context, and ask how specific sequences can condition states’ behavior. This extension also speaks to my interest in how “history matters.” For instance, does the order in which past settlement attempts occur affect the probability of future settlement attempts? The type of settlement attempt? How long the attempts last? The success of these attempts? Sequencing necessarily implies something about interdependence among a dispute’s settlement attempts. This interdependence is not a mere statistical nuisance to correct, but is substantively interesting if we are interested in assessing sequencing’s effect. We need a statistical estimator capable of modeling this specific kind of interdependence. We propose a new way of assessing questions about sequencing, in which we consider the entire sequence together, instead of examining at the individual events composing the sequence.

“Event Sequences in Disputed Issues”

Abstract
Do event sequences matter in territorial disputes between states? Existing work focuses on events’ shorter-term impact, both theoretically and empirically, but does not consider the events’ possible longer-term effects. Building from the literature on critical junctures, we articulate three conceptual requirements any argument must satisfy to make a compelling case for longer-term effects. We then argue the first settlement attempt’s type (peaceful or militarized) has the potential to meet these conditions by affecting disputant states’ propensity to use different types of settlement attempts in the future. We analyze Huth and Allee’s (2002) and the Issue Correlates of War’s (Hensel 2001) territorial dispute datasets to assess our hypothesis. We innovate by considering the entire event sequence, instead of examining the individual events composing the sequence, as current work does. We find suggestive evidence of longer-term first-event effects, contributing to our understanding of territorial disputes.

“Order? Order!”

Abstract
Does sequencing ever matter in political processes? Arguments involving event sequences, framing and priming, critical junctures, ballot ordering, and policy adoption (among others) suggest it does. However, for situations where we have only observational data available, we currently lack a good way to test inferences about ordering. Current methods rely on assuming conditional independence, or they examine observed sequences only, without reference to sequences that could have also occurred. We develop a generalized sequence model (GSM) to fill this need. Our innovation is using graph theory to construct an estimator capable of addressing the interdependence implied by sequencing, as well as using information about unobserved, but possible, sequences. We run simulations to demonstrate our estimator’s ability to recover sequence-related effects, and also compare our estimator’s performance to current alternative modeling methods in political science. We conclude with three brief applications, showcasing some of the sequence-related hypotheses our GSM is capable of assessing.