violence to other empirical regularities.
6
Conclusions
In this paper I have outlined some elements of a new model of voting under uncertainty.
This model is based on softer and empirically sounder assumptions than current models. Not
surprisingly the model performs better than current models when faced with already available
empirical regularities. One main example are the incentives of candidates to be perceived with
positive levels of ambiguity when they hold positions far from the position of the voter. In
addition the model is shown to be able to incorporate so-called abstention due to alienation as
a special case of abstention due to indifference and thus as rational and perfectly predictable
behavior. No current model couples these predictions with the observation that candidates
benefit from precision if they hold similar positions to those of the voter.
Another virtue novel to this formalization is to derive the importance and impact of the
different electoral dimensions from the factors that common sense points as relevant. Con-
trary to current formal models, here the weight of each dimension depends not only on how
important the voter thinks the particular dimension is and how much can she differentiate the
positions of the candidates but also how well she knows which position will best promote her
own interests.
The results are encouraging enough to address further tests and implementations of this
model in the future. I plan to derive and estimate the parameters and compare the per-
formance of the model presented here with that of its main alternatives against new data.
A straightforward test can be given by the 1996 National Electoral Studies used by Alvarez
(1998), which includes individual level data of all the relevant parameters of the model, such
as individual-level self-reported placement of their own position and that of the candidates in
a typical set of issue dimensions and the uncertainty they attach to these placements. Other
tests and applications can be developed for cross-country and cross-temporal analysis. Once
the parameters defining the electorates are measured party position equilibria can be obtained
for each election under study. Hopefully such a task will take us a little bit closer to under-
stand the electoral process and properly measure the incentives and elasticities faced by its
9
Convention