party, simply for controlling the presidency. Finally, as noted earlier, midterm elections are
viewed as referenda on the president and his administration. Those approving the president’s
performance in office support congressional candidates of his party while those disapproving
cast their votes for opposing party candidates. While the reasons for midterm losses by the
president’s party are debated, analyses of both macro-level election outcomes (e.g., Campbell
1997; Coleman 1997; Jacobson and Kernell 1983; Marra and Ostrom 1989; Oppenheimer,
Stimson, and Waterman 1986; Tufte 1975) and micro-level voter decisions (e.g., Abramowitz
1984, 1985, 1995, 2001; Born 1986; Jacobson and Kernell 1990; McAdams and Johannes 1984,
Piereson 1975; Ragsdale 1980) indicate that perception of the president is a significant a factor in
midterm congressional contests.
It remains unclear why assessments of the president’s job performance are so influential
in midterm congressional elections. One possibility is that assessment of the president provides
a convenient voting cue. Campaigns for Congress do not generate the public or media interest
that accompany presidential campaigns, leaving voters with less information about the
candidates and their positions. This lack of information about the candidates, or an
unwillingness to invest substantial resources in obtaining information, leads voters to use
shortcuts, basing their judgments between candidates on the information that is readily available
(Popkin 1991; Vavreck 2001). Voters might not know much about the congressional candidates
but they know how they feel about the president and the president’s policies and therefore use
this as a cue when choosing between congressional candidates.
Another possible explanation for the influence of presidential assessments in midterm
congressional elections is that voting is the product of displaced anger. Tufte (1975,813) argues
that with “no other targets available at the midterm, it is not unreasonable to expect that some
4
Convention