Shortly after Al Gore conceded defeat in the 2000 presidential election, he and President Clinton reportedly had a contentious meeting during which each blamed the other for the Democratic Party's loss of the White House. Given the strength of the U.S. economy and the president's high job approval ratings, Clinton naturally felt that he had created a positive political environment for Gore to succeed him. Gore's reluctance to involve Clinton in the 2000 campaign and to invoke the administration's record in his public appeals apparently irked the president. Clinton understandably felt he could have turned the tide, if only Gore had let him out to campaign. Of course, Gore did not see it that way at all. In his view, the president had become a liability as the result of the Lewinsky scandal. Had it not been for the scandal, Gore purportedly felt that he would have won with some margin to spare. He therefore felt wholly justified in keeping the president under wraps during the campaign. In sum, whereas Clinton felt that Gore would have won had he only taken greater advantage of their partnership, Gore felt that Clinton's indiscretions were a crucial liability that he could not escape from, as much as he tried.
This debate has consequences not only for interpreting the 2000 election result but also for understanding why all the political science models incorrectly predicted a clear Gore victory. If Clinton's spin on the outcome is correct, then the reason the scholarly models did not work in 2000 is that Gore ran a poorly focused campaign that failed to focus on the advantages that had led to victory in past elections. The broader message one could infer from such a conclusion is that campaigns really do matter. However, if Gore's take on the question is shown to be well founded, then the incorrect scholarly predictions were due to problems with the models themselves as opposed to strategic campaign decisions.
When the campaign managers met for a postelection conference at the University of Pennsylvania, they expressed a clear opinion that the political science models had missed a key variable. Matthew Dowd, a pollster for Bush, argued that because political scientists did not have time-series data on personal favorability, they were too reliant on presidential job approval ratings. As he put it, "When they premised the model on job approval rating without taking into account this personal favorability, which basically said, `It's time for a change because we're tired of the moral values and all that stuff,' they missed that calculation" (Jamieson and Waldman 2001, 25). Stanley Greenberg, a pollster for Gore, agreed that there were two distinct aspects to public evaluations of Clinton--job approval and personal favorability--and he noted that his analysis revealed they had an equal impact on voting decisions in 2000 (Jamieson and Waldman 2001, 91-92). Furthermore, his statewide polls showed that Clinton's personal ratings were particularly low in the key battleground states. Thus, like his Republican counterpart, he clearly felt that any academic model that included presidential job approval but not personal favorability would produce an overly optimistic prediction for Gore.
Although publicly available data on presidential favorability were lacking when academic prognosticators had to make their estimates, the National Election Study (NES) data now enable us to assess whether this variable was as crucial in 2000 as the pollsters maintained. There is a variety of ways to assess the public's views of Clinton based on the 2000 NES data, and because of the repeated use of identical questions in the NES over many years, it is possible to put these views into historical perspective. Despite the fact that Clinton's job approval rating was roughly equal to that of Eisenhower in 1960 and Reagan in 1988, public images of him were substantially more negative, right in line with what the pollsters suggested. …