3
election model. Assuming the measure does have explanatory power in terms of predicting election
outcomes, the remainder of the paper will focus on whether incumbents can use pork barreling, or
distributive spending, to ensure they face weaker campaigns in the general election.
Candidate Quality: Taking the Good with the Bad
When studying congressional elections, the literature has focused both on measures of quality and
candidate resources, specifically the ability to spend money. With respect to deterrence, however, the
preference appears to be developing a better measure of the quality of the candidates. There are both
substantive and empirical benefits to focusing on candidate-specific characteristics. Candidate quality,
first and foremost, should have a direct effect on the outcomes of elections. In fact, quality should not
only affect the outcome, but is also usually correlated with other common predictors of election outcomes
like incumbent or challenger spending. To not include some measure of quality leaves analyses open to
omitted variable bias and prevents a complete examination of elections. Another empirical benefit to
most measures of quality is that they are usually constructed in such a way that they are exogenous to the
system. This is not to imply that quality is uncorrelated with other predictors of vote share, but that these
predictors do not cause quality.
Quality, theoretically, is unobserved and there have been some notable attempts to craft a more
comprehensive measure of quality, both of the challenger and incumbent. One of the more recent and
more advanced measures comes from Stone and Maisel (2003) and Stone, Maisel, and Maestas (2004).
Using their Candidate Emergence Study they examine three dimensions of quality for the incumbent and
two for potential candidates (strategic and personal qualities for both and performance assessments for
incumbents). The measures here are based on assessments from participants and thus maintain the
attractive quality of exogeneity. McCurley and Mondak (1995) and Mondak (1995) also have a
multidimensional measure of quality, focusing on the quality of the incumbent. Their measures, also
derived from individual respondents, focus on the competence and integrity of the incumbent. Again, the
use of survey responses seems to contribute to the exogeneity of the measures. Other uses of survey
Convention