The Demographics of the 2004 Vote

It's much too early to forecast the eventual outcome of next year's election, but it's never too early to begin analyzing the way the election is shaping up.

George W. Bush has announced that he will run for re-election and is well underway with his major fundraising activities, but his Democratic opponent is ultimately still unknown at this point. There are at least nine Democratic candidates vying for their party's nomination and Gallup's analysis of historical patterns and current trial heat polling suggest that almost any of the Democratic candidates could eventually be nominated next July at the Democratic convention in Boston.

Thus, Gallup's approach so far this year has been to pit George W. Bush against "the Democratic candidate" in its research, a procedure that functions mainly as a generic re-elect measure and allows us to track the president's re-electability.

There has been some movement on the measure, with Bush hitting a high point of a 56% to 31% gap over the Democratic candidate back in October 2002, to a low point of 45% for Bush and 42% for the Democratic candidate on the eve of war in Iraq in March 2003.

The most recent Gallup Poll in which this measure was included was conducted July 25-27, and Bush's margin climbed slightly to 48% to 40% among the total adult population (and a 49% to 40% margin among registered voters).

But the focus of this article is the pattern of support for Bush and for his (as yet unknown) opponent.

It is obvious to all observers that Republicans will vote overwhelmingly for their party's candidate and Democrats for theirs, and that independents will end up being somewhere in the middle. The current partisan alignment in the United States is roughly one-third Republican, one-third independent, and one-third Democrat. Bush's lead comes about because Republicans are slightly stronger in their support for Bush than Democrats are in their opposition and because independents break for Bush rather than the Democrats.

The table below shows the relative ordering of support for Bush over the Democratic candidate within various combinations of the population based on party affiliation and ideology. This table is based on three different Gallup Poll samples*, for a combined total of 3,394 interviews. Across these three polls -- conducted June 12-18, 2003, July 18-20, 2003, and July 25-27, 2003 -- the overall breakout of projected voting for this large combined sample is 49% for Bush and 39% for the Democratic candidate. (Note that these data are among all national adults 18 and older, whether or not they are registered to vote at this point.)

It is intriguing to look beyond party identification, and identify the patterns of support for Bush versus the Democratic candidate based on demographic and geographic characteristics of the electorate.

The table below ignores partisan and ideological variables, leaving the list of voting intentions broken out by major demographic and geographic subgroups based on age, gender, education, income, race, religious behavior, and region of the country.

Here are the key points:

• Race is highly significant. Blacks are overwhelmingly likely to say they will vote for the Democratic candidate for president and whites say they will vote for Bush. This is not surprising. Only about 9% of blacks were estimated to have voted for George W. Bush in the 2000 election, and the results suggest a continuation of the same pattern.
  
  Hispanics at this point are also leaning toward the Democratic candidate by 47% to 42% -- a five-point margin over Bush -- although this is nowhere near the same level as found among blacks.
  
  
• Gender is also a highly significant variable in relationship to vote intention, although it is mediated by other characteristics such as age, education, and religious behavior.
  
  White women favor Bush over the Democratic candidate by 50% to 38%, while white men favor Bush by 60% to 30%. The gender gap, then, is 18 points -- comparing Bush's margin of 30 points among white men to his margin of 12 points among white women.
  
  There is virtually no difference in the voting preferences of black men and black women. Both groups overwhelmingly support the Democratic candidate.
  
  
• The major difference in vote intention by education comes between white voters who have postgraduate educations and those who do not. Both white men and white women who have education beyond the college level are significantly above average in their support for the Democratic candidate for president.
  
  White women with postgraduate degrees favor the Democratic candidate over Bush by a 20-percentage-point margin, while men with postgraduate education favor Bush by only a slim six-point margin. More generally, whites with postgraduate educations (regardless of gender) favor the Democratic candidate over Bush by seven points. In sharp contrast, whites who have graduated from college but have no postgraduate degree favor Bush by a 28-point margin.
  
  
• Church attendance is significantly related to vote choice, with those white Americans who attend church regularly much more likely to support Bush than those who attend church less frequently.
  
  The most extreme example of this tendency comes from a comparison of those whites who attend church every week with those who say they never attend church. Bush wins among the former by a 44-point margin, 67% to 24%, while he loses among the latter by a one-point margin.
  
  The group identified in this research with the highest level of support of all for Bush is white males who attend church weekly -- coming in with a 71% to 20% preference for Bush over the Democratic candidate.
  
  
• There are regional differences in white support for Bush. His strongest support comes among white Southerners, among whom he wins by a margin of 39 points. His weakest geographic region of the county is the West, where he wins by only a 12-point margin among whites.

• Bush has weak support among whites with the lowest incomes.

*The results are based on telephone interviews with a randomly selected national sample of 3,394 adults, aged 18 and older, compiled from polls conducted June 12-18, 2003; July 18-20, 2003; and July 25-27, 2003. For results based on this sample, one can say with 95% confidence that the maximum error attributable to sampling and other random effects is ±2 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.