Divided government affects individual choices over how to vote in midterm elections because it increases uncertainty in the minds of voters. Particularly, divided control of government makes blame attribution more difficult by obscuring causal connections and reducing the overall amount of usable information. As a result, we argue that under divided government, voters are less likely to vote for the House candidate not of the Presidents party. Using both NES and election-specific contextual data, we examine divided government's effect on the voters' political knowledge and candidate preferences in all midterm elections from 1978 to 1994, controlling for well identified factors that shape outcomes in House elections. We find, first, that divided government reduces the amount of political information held by voters. Second, divided government helps the President's party by lowering the probability that an individual votes for the out-party candidate.
The 1994 election resulted in one of the more dramatic changes of political fortune. in two years, the Democrats went from winning their first presidential election in 16 years to their largest midterm loss and, more importantly, the loss of control of the House of Representatives for the first time in 40 years. Was 1994 just simply the result of the known cyclical patterns in American elections or was something new at work?
The events of 1994 are consistent with some existing theories of midterm losses, inconsistent with others, and open to endless interpretation.1 It is not our intention to add to this. Rather, we wish to examine the 1994 election in its political context, comparing it to other midterm elections of similar and different circumstances. Specifically, we want to examine the choices in 1994 as they were perceived by the voters. What information could voters have about the performance of government, the positions of parties, and the effects of both on their lives? How does this level of information and uncertainty compare with the choices faced by voters in other midterm elections? What factors best explain these differences?
We hypothesize that voters in 1994 differed from recent previous midterm election voters in that they made their evaluations and arrived at their decisions on how to vote in the context of unified government. We believe that divided government obscures causal connections and increases uncertainty so that the overall propensity to vote against the candidate of the President's party is lower. We will show how voters' knowledge of partisan government controls changes across periods of unified and divided government and, further, demonstrate that these changes affect the propensity of voters to punish the President's party In so doing, we examine the preferences and decisions of voters in all midterm elections from 1978 to 1994, controlling for other well-identified factors that shape outcomes in House elections. Finally, we offer some thoughts on the implications of these findings to issues of electoral stability and governance.
THE IMPACT OF DIVIDED GOVERNMENT ON VOTERS
Studies of divided government fall into two broad categories: causes and consequences. Scholars interested in causes ask how and why voters produce divided control of the Presidency and Congress. Some look to how elites, both intentionally and unintentionally, structure vote choices that produce divided government (Alvarez and Schousen 1993; Jacobson 1990a; Petrocik 1991; Segura and Nicholson 1995; Wattenburg 1991). Others argue that voters intentionally choose divided government (Alesina and Rosenthal 1995; Fiorina 1992).
In contrast, research on the consequences of divided government explores its impact on public policy (Cox and McCubbins 1991; Fiorina 1992; Mayhew 1991; McCubbins 1991) and the strategic behavior of politicians in the legislative arena (Ginsberg and Shefter 1990; Kernell 1991). Curiously, this research has seldom asked how unified or divided government affects political attitudes among the citizenry The only two exceptions we know of are Bennett and Bennett (1993), who examine the factual information held by voters over time, and Leyden and Borrelli (1995), who examine state-level economic evaluations as endogenous to state-level divided government. …