Search
Browse By Day
Browse By Time
Browse By Person
Browse By Mini-Conference
Browse By Division
Browse By Session or Event Type
Browse Sessions by Fields of Interest
Browse Papers by Fields of Interest
Search Tips
Virtual Exhibit Hall
Change Preferences / Time Zone
Sign In
X (Twitter)
Students of voting behavior have not reached a consensus on the best practice to measure vote choice. While most employ survey respondents’ self-reported vote choices, others strongly advocate for using their likelihood of ``ever voting'' for each party, the so-called ``propensity to vote'' (PTV) scores.
In this article, we make a case against treating PTV scores as substitutes of self-reported vote choices. Our argument is based on the distinction between election-specific and longer-term determinants of electoral behavior. We argue that PTV questions, by design, are not intended to capture the former.
Applying a novel estimator, ``constrained choice conditional logistic regression'' (Moral and Zhirnov 2018, AJPS), to survey data from the 1994 Dutch Parliamentary Election Study and the 2002 Irish National Election Study, we model the determinants of individual voters’ vote propensities by examining their long- and short-term components seperately, and compare our estimates of those two types of propensities against individuals' responses to the PTV questions.
Informing our expectations from the two major schools of thought, we parameterize the long-term component using partisanship and ideological attitudes, and whereas the second equation on vote decision takes account of parties' election-specific issue positions, electoral viability, and campaign activities, and voters’ evaluations of party candidates/leaders and governments' economic performances.
The empirical analyses suggest that while PTV scores are highly correlated with our predicted probabilities of inclusion of the related parties in voters' individual choice sets, they have a weaker association with vote choice. We conclude by discussing the methodological problems that might arise from the use of PTV questions as the dependent variable, and suggesting that PTV and self-reported vote choice questions should be used as complements, rather than as substitutes.