Who's ahead? A close race decided by assumptions

Source: Scripps Howard News Service
Date: October 14, 2004

Democrat John Kerry leads President Bush by 5 percentage points, yet Bush is ahead of Kerry by 4 points.

Both statements are true even though they're based on the same poll.

The latest survey conducted at Ohio University's Scripps Survey Research Center offers a rare glimpse into the extremely close 2004 presidential race and the impact that different assumptions about likely voters have on survey results.

Unlike most other polls that report just one set of figures on the presidential race, this survey of 1,022 adult residents of the United States provides so-called "horse-race" data for Kerry and Bush based on a variety of methods. These somewhat conflicting findings offer insight into the mechanics of public opinion research.

The most elementary kind of poll is a household survey in which adults are interviewed by telephone after they are randomly selected. Researchers usually begin at this level since it makes the fewest assumptions about who should be interviewed, although household polls include people who are not voters, not citizens or even not living legally in the United States.

At this broadest level of public opinion research, President Bush is doing poorly. Only 43 percent approve of the job he has done as president and 53 percent say it's "time for someone new" when asked, "Would you like to see President Bush be re-elected to a second term?"

Among this vast group, Kerry is trouncing Bush 50 percent to 44 percent. (Six percent are either undecided or support a third party candidate like Ralph Nader.)

But registered voters are an important and quite different subgroup of America's residents. Poll respondents were asked: "Are you currently registered to vote at the address where you now live?" One out of seven respondents answered "No" to this. Registered voters tend to be older than average, are more likely to have children and more likely to own their own homes.

Among 867 registered voters in the survey, the race tightens slightly with Kerry at 50 percent and Bush at 45 percent. From this point on, the 155 people who said they are not currently registered will be ignored.

An important way to determine who's likely to show up at polling places on Nov. 2 is to ask how involved registered voters have been in past elections. The registered voters were asked: "How often would you say that you vote in major elections for president, governor or U.S. senator?" Seventy-two percent said "always," sixteen percent said "almost always."

When looking only at people who say they "always" or "almost always" vote, the race changes. Bush now leads with 48 percent to Kerry's 47 percent among people with a history of voting.

Millions of Americans are occasional voters. They will participate in elections when they are unhappy with the direction America is taking. And there is a definite unease about America's war on terrorism, its military operations in Iraq, the economy and vexing domestic problems like health care and the solvency of the Social Security program.

The poll asked: "Do you think America basically is headed in the right direction or in the wrong direction?" For many months, the Scripps Center has tracked a broad unease on this question. In the current poll, 46 percent say the nation is on the wrong path and 42 percent say it's headed in the right direction.

Sixty percent of registered voters who say they "rarely" vote in major elections also say America is headed in the wrong direction. If they go to the polls next month, they will support Kerry over Bush by more than a 2-to-1 margin.

These marginal voters are often ignored by pollsters because, by their own admission, they usually aren't players in the political process. But "rarely" does not mean never, and there are plenty of signs that marginal voters are taking interest in this election. In fact, that's what they're saying.

Voters were asked: "How much attention would you say you have been giving to the presidential election. Would you say you have been following the race very closely, somewhat closely or not too closely?" About 90 percent of regular voters say they are either "very closely" or "somewhat closely" watching the race. But about 75 percent _ an extraordinarily high number _ of registered voters who "rarely" vote in major elections say they also are watching the presidential race closely.

Another way to gauge how likely a voter is to cast a ballot is by asking how much passion he or she feels for the candidates. People in the poll who expressed an intention of supporting either Bush or Kerry were asked: "How certain are you that you will vote for this man?"

Bush has a slight advantage over Kerry when it comes to the depth of feeling in his supporters. Eighty-nine percent of voters who say they support Bush also say they are "very certain" they will vote for him, compared to 80 percent of Kerry's supporters.

If more stringent assumptions are made on who is certain to vote, Bush takes the lead over his challenger Among people who say they are "very certain" of their vote and who "always" participate in major elections, Bush leads Kerry 52 percent to 48 percent.

The survey was conducted Oct. 3-13 at Ohio University's Scripps Survey Research Center in Athens, Ohio. The project was funded in a grant from the Scripps Foundation.

A survey of this type has a margin of error of about 4 percentage points when looking at all participants in the poll. The margin rises, however, when subgroups within the poll are examined. The margin is nearly 6 percent for the most likely voter calculation.