The Neglected Role and Variability of Constants in the Spatial Valence Approach: An Assessment

24.09.2018 - 12:00 bis 13:30
A 5,6 Raum A 231
Art der Veranstaltung: 
AB B-Kolloquium
Dr. Ingrid Mauerer
Zugehörigkeit des Vortragenden: 
LMU München

Empirical applications of the spatial theory of elections typically rely on discrete choice models to arrive at operational probabilistic voting models. Whereas in classical spatial models voter choice is solely a function of spatial proximity, neo-Downsian models also incorporate voters’ nonpolicy considerations. One prominent line of probabilistic voting models, Schofield’s Valence Model, additionally includes party or candidate valences into voter utility functions. To measure the valence advantages empirically, the model rests on the estimated party intercepts, which are ordered based on their absolute values, and then this valence ranking is used further to predict equilibrium locations. The paper demonstrates that this measurement strategy is challenging due to central properties of discrete choice models and the specific nature of party intercepts. The latter represent the relative average role of all unobserved factors, measurement or specification errors. First, it illustrates that in fully-specified voter models only a small amount of unobserved utility sources should end up in the constants – that might or might not be nonspatial party qualities. Second, it demonstrates that the valence ranking, the key factor to investigate how valence differences affect the nature of spatial competition, is highly sensitive to arbitrary coding decisions. I use mass election surveys from Germany to demonstrate these points.