This research involves the second wave of a panel study of the American electorate begun with federal funds at the university of the researchers. The project identifies the distribution, stability and roots of segmented partisanship and assesses its impact on the way in which Americans think about and react to politics, political institutions, and candidates. The role of political parties in ordering democratic institutions and in framing electoral choice has not been seriously questioned since political scientist E.E. Schattschneider proclaimed, in 1942, that " . . . modern democracy is unthinkable save in terms of the parties." Similarly, at the individual level, while its predictive power has fluctuated over time, the centrality of party identification to any understanding of electoral turnout, vote direction, the assessment of candidates, the impact of issues, and the role of ideology has never been seriously disputed. If anything, the most productive research has enriched our appreciation of partisanship by demonstrating its multidimensionality. The impact of federalism on the dimensionality of partisanship, however, has been largely ignored. Research has emphasized the constancy of party identification within each citizen across levels of government. Because the scope and type of political decisions often vary among levels of government, Americans may develop different expectations for the performance of government at different levels. Hence, it is both logical and often appropriate for voter approval, contempt, and indifference to be localized and not fully generalizable across all levels of government. Moreover, when party programs and personalities are linked to level-based performance expectations, variation in citizen response to those programs and personalities should be an expected, not a surprising, outcome. In short, it should be anticipated that people will respond differently to the different powers and purposes of national and state government. This second wave of interviews in the panel study of the American electorate involves 1,350 respondents selected using random-digit dialing and random selection within households. It should allow greater understanding of the way in which Americans think about and react to politics, political institutions, and candidates.
