Source Material: Presidential Data Locator

Abstract

The Question of Approval This inaugural edition covers one troublesome data element: public approval. Does presidential public approval affect congressional support? The answer to this question constitutes a conundrum. To practitioners, the answer seems quite clear: Yes, certainly, why do you ask? When political scientists try to answer this question, their responses present a tumult. Empirical research has not consistently supported these arguments (see Collier and Sullivan 1995 for a review). While some studies have found evidence to support the impact of prestige (see Brace and Hinckley 1992; Edwards 1976, 1980, 1989, 1997; Ostrom and Simon 1985; Rivers and Rose 1985; Sullivan 1987), other studies have found little or no impact of public opinion on congressional support, especially when research focuses narrowly on just the administration's highest priorities (see Collier and Sullivan 1995; Lockerbie, Borrelli, and Hedger 1998; Cohen et al. 2000). Other research points to the complexity of the linkage between presidential popularity and congressional support, in which being popular leads to more support from the president's own party but in fact less support from the opposition (Bond and Fleisher 1980). Bond, Fleisher, and Wood (2002) suggest a theoretical rationale to account for approval's varying effects. As a profession, we are as unclear about the impact of public approval as the professionals are certain about its usefulness. One possible approach to explaining this conundrum goes to a measurement issue, challenging the use of national opinion polls, and is aided along by the release of a new data set. Two papers recently presented to the Midwest Political Science Association take this approach and use this new data. One, by a team of collaborators at Texas A&M and Fordham focused on senators and the electoral linkage assumed to drive the relationship between presidential leadership and congressional responsiveness (Bond et al. 2002). A second team of collaborators from the University of North Carolina at Chapel Hill, Minnesota, and Columbia focused on the House and whether more specific approval information affected the conversion of otherwise nonsupportive members (Hora et al. 2002). When we assume that legislative politics begins at the local level and never departs, reliance on national opinion thermometers, devoid of local contexts, seems a significant operational mistake. Until recently, we had little else to use but national opinion surveys. (1) Job Approval Ratings For decades, local organizations, mostly newspapers, have collected public opinion information focused on narrowly defined constituencies. A recent National Science Foundation-funded project at three universities began collecting and collating these statewide and sometimes city- or congressional district-sized public opinion surveys. While the bulk of the data (arranged in three data collections) focuses on governors, one data set presents approval ratings for presidents. The Job Approval Ratings OARS) database ranges over the core of the American century, Franklin Delano Roosevelt onward through Clinton, but presidential approval ratings concentrate mainly on the last three presidents. For example, Figure 1 presents the density of state-level polls for New York and Tennessee, two important states for the Clinton-Gore administration. Juxtapose that density with the fact that the data set holds only one state-level poll for New York during the Johnson administration. For the last three administrations, the database holds results from about 1,200 polls. For questions, like the Bond et al. (2002) examination of the electoral connection, this data set has a good deal to offer. [FIGURE 1 OMITTED] Minor Quibbles It does also have some problems. For example, the state codes in the data do not match the venerable Inter-university Consortium for Political and Social Research (ICPSR) state codes that govern all congressional data sets. …