A better showing among movable voters has helped put George W. Bush back at the 50 percent mark, with 47 percent support for John Kerry among likely voters as the 2004 presidential campaign enters its closing days.
The race is still very close: Given polling tolerances, a three-point gap is not large enough to be conclusively characterized as a lead. Bush has improved in the last few days among some groups, such as movables, conservative Democrats and young adults. But Kerry's doing a bit better among white Catholics, a key swing group, helping to keep it tight.
The ABC News tracking poll shows no impact from the controversy over missing munitions in Iraq. Importance of Iraq as a voting issue hasn't significantly changed; Kerry still leads broadly -- but no more so -- among those who call it their top issue.
This poll is based on interviews Monday through Thursday; results from last Saturday and Sunday, which were two of Kerry's three best days since this tracking poll began Oct. 1, have rolled out of the average. The results the past four days have been quite stable -- a small numerical advantage for Bush, within the margin of error, each night.
The race has been quite stable among voters who say they've definitely made up their minds (essentially an even split between Bush and Kerry). There's been more change among movable voters -- those who haven't made a definite choice.
Movables in today's results favor Bush by 52-33 percent, a change from 46-36 percent in Kerry's favor as of Sunday. It's a small group -- eight percent of likely voters, down from 14 percent when tracking began -- but in a close race, an important one.
Questions about movables remain. One is whether they keep moving; they still, by their own description, have not definitely decided. Another is how many of them actually vote: Movables are 30 points less likely than decided voters to be following the race very closely; that suggests less commitment to the process, which may mean lower turnout in this group.
Movables also are less partisan, more ideologically moderate, and younger -- about one in four are under age 30. Non-partisan and young voters can be lower-turnout groups.
The two traditionally key swing voter groups, meanwhile, have returned to about even splits: Independents divide 48-47 percent and white Catholics 49-47 percent for Kerry vs. Bush. That's better for Kerry among white Catholics, and better for Bush among independents, than early in the week.
In every election since 1980 (all those for which exit polls are available) the candidate who has won both independents and white Catholics has won the presidency.
Youth and Marriage
Three other important groups -- whose preferences have not followed their 2000 pattern -- are young likely voters, unmarried men and married women.
Young voters, age 18 to 29, have shifted this week from 60-35 percent Kerry-Bush to a closer 52-45 percent (and as noted, movables are disproportionately young). That's still better for the Democrat than in 2000, when young voters split evenly, 48-46 percent, between Al Gore and Bush.
Young voters have been a key to Kerry's competitiveness; the latest results indicate that his challenges include not only encouraging their turnout, but also holding their loyalty.
Kerry's lead among unmarried men, similarly, has moderated from 63-23 percent Sunday to 56-41 percent now. That is still a significant lead, and enough to help keep Kerry in the hunt; single men in 2000 divided 48-46 percent between Gore and Bush.
While Kerry continues to do better than Gore did among single men, Bush is doing better among married women -- 55-43 percent over Kerry. Married women in 2000 split evenly.
Bush also is back to poaching more from his opponent's side: Twelve percent of Democratic likely voters support Bush, while six percent of Republicans prefer Kerry. Each party's candidate usually loses some from his side; the aim is not to let it happen disproportionately.
A difficulty for Kerry is that conservatives account for 15 percent of all Democrats, while liberals account for fewer, five percent, of all Republicans. And indeed, the Democrats who support Bush are much more likely to be conservative Democrats (and much less likely to be liberals).
Democrats who favor Bush also are more likely to be white (eight in 10, compared with two-thirds of all other Democrats); higher-income (58 percent are in $50,000+ households, compared with 44 percent of other Democrats); and Southerners.
Other groups are dividing much a they have in the past. Among racial groups, whites are dividing by 57-40 percent for Bush; blacks, 92-6 percent for Kerry; Hispanics, 68-28 percent for Kerry. And Kerry leads solidly among lower-income voters, while Bush is strongest among the better-off.
This survey also finds that 12 percent of "likely voters" in fact already have voted. As noted previously, they're disproportionately Westerners, older, somewhat more apt to be Republicans and tilt toward Bush.
Preference on the issues have been fairly stable lately: In the top three, 24 percent now call Iraq the single most important issue in their vote, 23 percent say it's the economy and jobs and 20 percent say it's terrorism.
Terrorism peaked at 28 percent after the Republican convention; it's decline has made it a tighter race, because those who pick it favor Bush by an overwhelming margin, now 88-11 percent. Those who call the economy the top issue favor Kerry by 71-25 percent; those who say it's Iraq favor Kerry by 59-38 percent.
Kerry has a similar lead among those who call health care the most important issue. And the lead goes back to Bush, by 68-29 percent, among those who cite some other issue as No. 1 -- 18 percent of all likely voters, with a wide range of factors and attributes motivating their ultimate choice.
This poll was conducted October 25-28 among a random national sample of 2,820 adults, including 2,488 registered voters and 2,047 likely voters. The results have a two-point error margin for the likely voter sample. ABC News and The Washington Post are sharing data collection for this tracking poll, then independently applying their own models to arrive at likely voter estimates. Sampling, data collection and tabulation by TNS of Horsham, Pa.
Click here for PDF version with full questionnaire and results.
See previous analyses in our Poll Vault.