In this article we analyze the campaign behavior of presidents in the congressional midterm elections held between 1954 and 1998. We seek to identify those factors that influence a president's decision to visit a particular state during the midterm election season and campaign. Employing a negative binomial regression model, we find that presidents determine their midterm election campaign schedule based on a consideration of the differences among the 50 states in terms of both the chief executives personal popularity and the number of competitive races being held. The importance of these factors suggests that there is an underlying strategy to a president's campaign activity during the midterm election season that is comparable to campaigns for the presidency itself.
In this article we seek to explain the midterm election campaign behavior of presidents between the years 1954 and 1998. We model the factors that determine where a president chooses to campaign among the fifty states during a midterm election season given both the limited time and finite resources at his disposal. We find that the midterm election campaign behavior of presidents is based on a strategic calculation designed to maximize the number of his copartisans found in both the national Congress and governorships of the 50 states.
There has been a surprising lack of previous inquiries into this aspect of American politics. Past research on the issue of presidential involvement in midterm election campaigns has been limited to considering the effects of presidential campaign visits on election results (Cohen, Krassa, and Hamman 1991).' We believe that our study is the first to offer a systematic explanation for this aspect of presidential decision-making and behavior.
PRESIDENTS AND MIDTERM ELECTION CAMPAIGN STRATEGY
In contrast to the lack of work investigating the midterm election behavior ol presidents, numerous studies analyze campaigns for the presidency itself. These works share a common view that candidates for president strategically determine their campaign schedules to maximize vote totals in the Electoral College and thus improve their prospects (or election (Colantoni, Levesque, and Ordeshook 1975; West 1983; Bartels 1985; Gurian 1986, 1993; Rabinowitz and MacDonald 1986).
One could plausibly argue that a president is less likely to engage in such a strategic calculation when deciding on campaign stops lor a midterm election as compared to those elections deciding the presidency. The president's own position is unchallenged regardless of the election results and so his personal stake in the outcome is less vital. The president may also perceive the election as an opportunity to reward friends and contributors with a campaign visit as opposed to developing a broader electoral plan designed to benefit his political party or legislative priorities. If this viewpoint proved accurate, we would expect there to be no predictable pattern to presidential campaigning in anticipation of midterm elections.2
We seek to demonstrate that presidents do in fact develop a strategy for participating in the midterm election campaign season that is comparable to the one employed for elections to the presidency itself. But if a strategic calculation in determining the president's campaign schedule for midterm elections exists, it is of necessity informed by a different set of concerns than those associated with presidential elections. During the midterm election campaign, presidents are chiefly preoccupied with ensuring a favorable configuration of the Congress that will make the Constitution's "invitation to struggle" between the executive and legislative branches less arduous. This can be accomplished by electing as many government officials as possible that share the party affiliation and ideology of the current president.3
Campaign visits comprise one significant resource at the president's disposal to encourage a favorable configuration of the future Congress. …