Aug. 6, 2004 Alan I. Abramowitz
When Good Forecasts Go Bad: The TimeforChange Model and the 2004 Presidential Election Despite mediocre approval ratings and below average economic growth, George W. Bush has a good chance of winning a second term in the White House according to the timeforchange forecasting model. Based on the assumption that a presidential election is fundamentally a referendum on the performance of the incumbent president, the timeforchange model uses three variables to predict the outcomes of presidential elections more than three months before Election Day: the incumbent president’s approval rating in the final Gallup Poll in June, the change in real gross domestic product during the first two quarters of the election year, and a dummy variable based on whether the president’s party has controlled the White House for only one term or for more than one term. This final variable, the timeforchange factor, plays a critical role in the model because a firstterm incumbent like George W. Bush generally enjoy a large advantage compared with a candidate whose party has held the White House for two terms or longer. Since 1900, only one firstterm incumbent, Jimmy Carter, has been defeated. The timeforchange model has correctly predicted the popular vote winner in every presidential election since 1988 (Abramowitz 1988; Campbell and Mann 1992; Abramowitz 1996; Abramowitz 2001). For all presidential elections between 1948 and 2000, the average outofsample forecasting error of the model was less than two percentage points. In 2000, the timeforchange model predicted that Al Gore would receive 53 percent of the major party vote. This was much closer to Gore’s actual vote share (50.2 percent) than the large majority of 2000 election forecasts, many of which predicted that Gore would win in a landslide (Campbell and Garand 2000). For the 2004 election I have made no changes in the basic timeforchange model. However, I have incorporated revised estimates of real GDP growth that were recently produced by the Bureau of Economic Analysis of the Department of Commerce. In addition, I have slightly modified the presidential approval measure. Although the Gallup Poll has been asking the same presidential approval question since the 1940s, in recent years Gallup has pressed respondents much harder to express an opinion. As a result, the percentage of respondents with no opinion about the president’s job performance has declined dramatically, from 1015 percent in earlier years to less than five percent recently. In order to correct for this change in Gallup’s procedures, the current version of the timeforchange model uses the president’s net approval rating which is the difference between his approval rating and his disapproval rating. Using these revised measures, I reestimated the timeforchange model with data from the 1948 through 1996 presidential elections and used the results to correct the original forecast of the 2000 election. The revised “forecast” of 50.3 percent of the major party vote for Gore came within onetenth of a percentage point of the actual results. Reestimating the timeforchange model with data for the 1948 through 2000 presidential elections produces the following results: V = 50.75 + .107 (.025)*JUNEAPP + .818 (.183)*FHGDP – 5.14 (1.24)*TERM, where V is the incumbent party’s predicted share of the major party vote, JUNEAPP is the difference between the president’s approval and disapproval ratings in the final Gallup Poll in June, FHGDP is the annualized growth rate of real GDP during the first two quarters of the year, and TERM is a dummy variable that takes on the value 0 if the president’s party has been in office for one term and 1 if the president’s party has been in office for more than one term. The standard error for each regression coefficient is shown in parentheses. All of these coefficients are highly statistically significant (p < .001) and the regression equation explains just over 90 percent of the variance in the outcomes of presidential elections since World War II. Based on these results, the timeforchange model predicts that George W. Bush will receive 53.7 percent of the major party vote in the 2004 presidential election. The standard error of this forecast is plus or minus 2.0 percentage points. George Bush’s net approval rating of 1 percent is well below the average of +11.7 percent for all incumbents running for reelection and far below the average of +19.4 percent for all firstterm incumbents. In addition, the estimated 3.75 percent growth rate of the U.S. economy during the first half of 2004 is below the average of 4.5 percent for all presidential election years since World War II. The timeforchange model predicts a Bush win in 2004 primarily because he is a firstterm incumbent. According to the results presented above, a first term incumbent typically receives a boost of just over five percentage points compared with a candidate whose party has controlled the White House for two or more terms. Despite this advantage, however, George Bush has been running no better than even in the polls with his Democratic challenger, Massachusetts Senator John Kerry. In 25 national polls conducted during the month of July, Bush was favored by an average of 44.5 percent of likely voters compared with 46.3 percent for Kerry and 3.4 percent for independent candidate Ralph Nader. It is entirely possible that George Bush will overcome his current deficit in the polls and go on to win the 2004 presidential election as the timeforchange model predicts. Given the stability of the race for the past five months, however, and the fact that Bush’s largest average lead in any month was just over one percentage point, it seems unlikely that he will receive anything close to 53.7 percent of the major party vote in November. The poll results suggest that George Bush may not receive the full advantage that normally accrues to an incumbent president when his party has been in office for only one term. As many commentators have noted, the U.S. electorate is exceptionally polarized in 2004. George Bush enjoys overwhelming support from his copartisans but he is not getting anywhere near the level of support that a firstterm incumbent typically gets from independents and opposing partisans. In fact, the difference between President Bush’s approval rating among Republicans and Democrats is the largest ever recorded by the Gallup Poll. This high level of polarization may reflect the disputed outcome of the 2000 election as well as intense divisions within the electorate over many of the Bush Administration’s policies including the war in Iraq. Whatever its causes, the current polarization of the U.S. electorate means the 2004 presidential election will probably be very close and the timeforchange model may chalk up its first erroneous prediction. 
