1
A number of studies have examined the electoral consequences of Congress members’
voting behavior; recently Canes-Wrone, Brady, and Cogan (2002) explore the effects of
spending, ideological extremity, and presidential popularity as well as other factors on the vote
share received by incumbent House members who faced challengers in the general election.
They find that incumbents who support the ideological extremes of their parties receive lower
vote shares and are less likely to be reelected than their more moderate colleagues.
In this paper we offer both methodological and theoretical refinements to the relationship
between incumbent ideology and electoral outcomes developed by Canes-Wrone et al (2002).
First, we argue that the exclusion of uncontested elections from their analysis may introduce
selection bias into their results. Their findings regarding the effect of ideological extremity may
be an artifact of using OLS in the presence of selection bias. In order to correct for possible
selectivity bias, we examine the effects of legislative behavior and candidate characteristics on
the electoral margins of U.S. House members using a selection bias model. The Heckman
selection model uses maximum likelihood techniques to estimate a system of equations. In our
model, the selection equation predicts the likelihood that an incumbent is contested in the general
election and is included in the substantive regression equation predicting the incumbent’s vote
share in the general election.
In addition to this methodological issue, the theoretical relationship between ideological
extremity in roll call voting and the electoral fortunes of Congress members can also be refined.
We argue that the electoral costs of roll call ideological extremity vary across members based on
their constituencies and personal characteristics. We expect that members who exhibit
ideological extremity in their roll call voting and represent ideologically extreme districts may
not pay the same price for their legislative behavior as members from more moderate districts; in