|
UPCOMING CONFERENCE PRESENTATIONS
"The Decision to Protest as a Constrained Choice Problem."
To be presented at the 113th Annual Conference of the American Political Science Association in San Francisco, CA. Abstract Previous literature on protest behavior has long focused on micro-level factors, such as grievances, resources and values. More recent studies add macro-level factors such as political and economic development (Dalton et al. 2010) or regime type (Bernhard and Karakoç 2007) to this list, arguing that political, economic, and institutional structures affect protest frequency and magnitude by providing individuals with opportunities. Previous comparative studies commonly employ additive indices without accounting for varying costs of and constraints on distinct forms of protest. In fact, protests of political nature range from democratic, peaceful means of revealing one's discontent with the government, such as signing a petition, to using more confrontational, even violent means. Additive measures and linear models employed in previous studies, however, do not allow us answer why individuals prefer one type of protest to another, and why particular types of protest are more common in some countries but not in others with similar levels of political and economic development, or with similarly democratic regimes. This study seeks to answer whether social, political and institutional contexts act as constraints on protest behavior and make certain types of political protest unavailable to some citizens. I employ a novel discrete-choice model, constrained choice conditional logistic regression (Moral and Zhirnov 2017), that allows researchers to model two interrelated processes: the process that generates an individual's effective choice set and the process whereby the individual makes a choice among the alternatives included in her choice set. Employing data from the fourth wave of the World Values Survey and examining a large set of established democracies with varying social, political and institutional structures, the propensity of all alternative forms of protest to be included in individuals' effective choice sets is modeled as a function of state capacity, legal constraints, and the availability of other less costly forms of revealing discontent. The decision to choose a particular form of protest, on the other hand, is modeled as a function of micro-level grievances, resources and values as robust determinant of protest behavior in previous literature. The empirical analyses show that higher state capacity, stricter legal constraints, and non-permissive electoral rules act as constraints on protest behavior and make both violent and non-violent means of political protest unavailable to some individuals regardless of their grievances, resources, and values. In addition to explaining why some well-established democracies such as France exhibit more frequent and more violent means of political protests which are seldom observed in others, this study contributes to the literature by showing that distinct social, institutional and political structures not only provide individuals with necessary opportunities, but also act as constraints on their protest behavior. "Ever Considered (Not) Voting for a Party?" With Andrei Y. Zhirnov. To be presented at the 113th Annual Conference of the American Political Science Association in San Francisco, CA. Abstract The choice of the dependent variable has long been a concern to the students of voting behavior. While some advocate using respondents' intended or actual vote choices and employing discrete-choice models, others prefer to use respondents' self-reported likelihoods of "ever voting" for each party, or the so-called propensity to vote (PTV) questions (as they appear in a number of national and cross national election studies since 1980s). The proponents of the latter approach contend that PTV questions provide researchers with a direct measure of the latent (electoral) utilities in discrete choice models (van der Eijk et al. 2006). We argue this is not the case. Such questions, by design, are not intended to capture the immediate consequences of specific campaigns and party policy offerings, which enter individuals' calculus of voting along with their longer-term considerations. The constrained choice conditional logistic regression (Moral and Zhirnov 2017) allows researchers to model the observed vote choice as a combination of two distinct processes: the process that generates an individual voter's effective choice set and the process whereby the voter makes a vote choice among the included party alternatives. Applying this novel model to the data from the Dutch Parliamentary and European Election Studies, we assess the validity of this concern and whether the PTV questions can be used in lieu of respondents' self-reported vote choices in empirical studies on spatial theories of voting. We model individuals' propensities to include parties in their effective choice sets as a function of some long-term factors: their partisan attachments and strategic considerations, and the extremity of party policy offerings. We parameterize the second equation using short-term, election-specific changes in the policy offerings of parties. The empirical analyses show that the PTVs are not equivalent to the latent utilities underlying observed vote choices. While PTVs are highly correlated with individuals' estimated propensities to include each party in their effective choice sets, they have indeed a weaker association with the effects of policy concerns on the observed vote choice. In the concluding section we discuss some methodological problems regarding the use of PTV questions as the dependent variable, and argue how PTV and vote choice questions should be treated as complements, rather than substitutes, in empirical investigations of voting behavior. |
REPLICATION DATA
Please click here for my dataverse where you can find the replication data and code for my published and forthcoming research.
|
