Key Insights From Election

GALLUP NEWS SERVICE

PRINCETON, NJ -- An analysis of Gallup's last pre-election poll suggests several lessons and insights about the 2004 presidential campaign.

1. The basic patterns of this election were very similar to 2000.

In only three states did a majority of voters switch their support from one party to another: New Hampshire voted for President George W. Bush in 2000, but for Sen. John Kerry in 2004; New Mexico and Iowa both supported Al Gore in 2000, but gave their electoral votes to Bush in 2004.

The popular vote in 2004 was also similar to the popular vote in 2000, according to an analysis of the final Gallup pre-election polls in each year. Kerry received the same percentage of the popular vote as did Gore in 2000 (48%), while Bush's percentage went up by three points (48% to 51%). Minor party and independent candidates went from 4% in 2000 to 1% this year (based on preliminary national popular vote figures).

Bush's percentage of the vote among most subgroups was generally proportional to the overall three-percentage point increase he received. For example, his percentage of the vote among men rose by four percentage points, while among women it rose by three points, suggesting essentially no significant difference in the gender gap this year from what was found in 2000.

As detailed in How Americans Voted, the main exceptions that show Bush receiving proportionately more support in 2004 than in 2000 are found among voters living in urban areas (Bush received nine percentage points higher support in 2004, compared with 2000), conservatives (+9), non-union households (+8), those who attend church weekly (+7), and Protestant households (+7). Bush lost support compared with 2000 among 18- to 29-year-olds (-7 percentage points), people living in rural areas (-6), moderates (-4), and unmarried men (-4).

But the overall pattern of Bush and Kerry's support is quite similar to 2000. Bush's major advantage came among non-Hispanic whites, men, 30- to 49-year-olds, those living in the South, those with some college or a college degree (but not those with post-graduate educations), married people, those who attend church frequently, and Protestants. Kerry's main support came from women (but this was not dramatic), nonwhites (blacks in particular), 18- to 29-year-olds, Easterners, urban dwellers, those with post-graduate degrees or no college, unmarried people, union households, those who seldom or never attend church, and Catholics.

2. The events that had the most impact on this campaign were the Republican National Convention and the first debate.

There are only five highly visible, structured events in a presidential campaign (once the nominees are known) that have the potential to shake up a presidential contest: the two party conventions and the three debates.

In this election, an analysis of the trend lines on support for the two candidates throughout the last five months of the campaign suggests that the most important of these structured events were the GOP convention held in New York in late August/early September, and the three debates taken as a whole.

Bush gained significantly in the trial heat polling of likely voters after the GOP convention, an improved positioning he held onto until the first debate in Miami. At that point, Kerry immediately moved back to parity with Bush, a position he held through the period of the three debates.

After the debates, however, Bush regained his lead over Kerry, a position he held until his victory on Election Day.

3. Kerry's emphasis on his Vietnam service did not produce gains for him in voters' perception of his leadership, nor of his ability to handle Iraq or terrorism.

Kerry and his strategists made the decision to put a heavy emphasis on Kerry's Vietnam service at the Democratic convention, in order to counter Bush's strength among voters on leadership, especially military leadership. But the polls suggest Kerry's strategy did not work. Kerry always trailed Bush by large margins on which candidate could better fight terrorism and which candidate was a strong and decisive leader.

The strategy may have failed in part because of the Swift Boat ads that attacked Kerry's credibility. But Gallup's polls show that even prior to the ads, the Democratic convention provided no boost to Kerry, one of the few times in recent elections when a party's convention did not provide a "bounce" in the polls for the party's presidential candidate.

4. Bush won this year "despite" several obstacles: despite having job approval under 50%, despite the fact that satisfaction was well below 50%, despite having a significant plurality saying the economy was getting worse, despite having only half the nation supporting his signature Iraq initiative, and despite his opponent beating him on almost every domestic issue.

Each presidential election is a new one, of course, and this is no exception. There have been only 10 elections now since World War II in which an incumbent has run for re-election, a small sample from which to make generalizations. Nevertheless, a number of key indicators from the population as a whole suggested throughout the year that Bush would not sail to an easy victory.

The sample of 10 post-World War presidents who ran for re-election shows that the incumbent's job approval rating has been overall a good predictor of re-election probabilities.

Gerald Ford with a 45% approval rating, Jimmy Carter with a 37% approval rating, and George H.W. Bush with a 33% approval rating all lost their re-election bids. Before 2004, every other president who won, except for Harry S. Truman, enjoyed an approval rating of 54% or better.

The pattern suggests that presidents who have very low job approval ratings are likely to lose, while those with job approval ratings above 50% are likely to win. However, both the current President Bush and Truman both won despite low approval ratings.

Other hurdles that Bush overcame in the election:

• In the last pre-election poll, voters said that Kerry, rather than Bush, would better handle the economy by 50% to 48%.
  
• Also in that poll, 58% of Americans said the Bush administration was responsible for the missing munitions in Iraq either a "great deal" (33%) or a "moderate amount" (25%).
  
• In an Oct. 22-24 poll, the public was evenly divided as to whether they agreed or disagreed with Bush on the issues that mattered most to them -- 49% said they agreed with Bush, 49% disagreed.
  
• Also in that poll, the public expressed greater confidence in Kerry than in Bush to handle healthcare issues, by 55% to 41%.
  
• More people saw Kerry than Bush as caring about the needs of people like them, by 51% to 43%.
  
• Bush's favorable ratings among the public were virtually identical to Kerry's. Among all Americans, 52% felt favorably about Kerry, 51% about Bush. Among likely voters, 53% were favorable about Bush, 51% about Kerry.

5. The likely voter model is a better estimate of the actual vote than results based on all registered voters.

There was an enormous amount of discussion coming into this election that there would be highly unusual patterns of voting, confounding the traditional methods of ascertaining likely voters in pre-election polls. Filmmaker Michael Moore and many others opined that newly registered voters and newly activated voters would confound pollsters and that Kerry would do much better in the final voting than what pre-election polls predicted.

Gallup's final pre-election poll conducted Oct. 29-31 showed the following:

These data suggest that had all registered voters actually turned out and voted on Election Day, Kerry would have won the popular vote by two percentage points. Gallup's model of the pool of likely voters, on the other hand, showed that those actually most likely to turn out on Election Day supported Bush over Kerry by a two-point margin. The preliminary tallies of the popular vote suggest that Bush defeated Kerry by a three-point margin, very close to what Gallup's likely voter model assumed.

Survey Methods

These results are based on telephone interviews with a randomly selected national sample of 1,573 likely voters, aged 18 and older, conducted Oct. 29-31. For results based on this sample, one can say with 95% confidence that the maximum error attributable to sampling and other random effects is ±3 percentage points. In addition to sampling error, question wording and practical difficulties in conducting surveys can introduce error or bias into the findings of public opinion polls.

The results are adjusted to reflect the overall national popular two-party vote.