But as pundits, prognosticators and pollsters have discovered, 2008 is far from a normal election year.

A new Dispatch Poll shows that the New York senator would lead in Ohio by as much as 16 percentage points under the usual turnout scenario dominated by loyal, older party voters -- especially if women show up in force, as they have in earlier primary states.

Even if you presume an unusual influx of young and black poll-goers, Clinton still would hold a healthy single-digit lead, the poll indicates.

So Tuesday's election likely comes down to how many independent voters decide to grab a Democratic ballot and vote for Illinois Sen. Barack Obama. Secretary of State Jennifer Brunner's prediction of a record 52 percent turnout means that well more than a quarter of Ohio's 5 million-plus nonpartisan voters will vote.

But there's one undeniable fact among Democrats: They overwhelmingly think that Obama will win the November election, regardless of whom they support in the primary. Even nearly a third of Clinton supporters in the poll expect Obama to be the next president.

One of them is Diana Brenner, a 54-year-old psychologist from Columbus, who said she likes Clinton's experience and the way she can specifically explain her positions on health care and the economy.

But Brenner still thinks Obama is going to win.

"This country is so tired, especially after the last seven years, of such a negative environment," she said. "He seems so new and fresh that I think people are being swayed in his direction because of that."

Ohio is generally considered a must-win "firewall" for Clinton, although her supporters are now saying she has to win only two of the remaining three major states. Texas also votes Tuesday, while Pennsylvania's primary isn't until April 22.

But because of complicated rules for selecting Democratic delegates, modest Clinton victories provide little help in narrowing Obama's overall lead.

Things are more set for Ohio Republicans. The only real question is Arizona Sen. John McCain's margin of victory; the poll shows him up 64 percent to 24 percent over former Arkansas Gov. Mike Huckabee.

Even though McCain appears on the verge of mathematically clinching the nomination, Ohio GOP voters are still unhappy with their choices. More than half proclaimed themselves "not very" or "not at all" satisfied with their party's presidential candidates.

All polls in Ohio show Obama rapidly narrowing a Clinton lead that stood at more than 20 points for months. But pollsters have a mixed record this year of sorting out the presidential race in several states, starting with New Hampshire.

"The big issue is that the media, and many pollsters, treat primary elections as if they are general elections," said Andrew Smith, director of the University of New Hampshire Survey Center who got burned forecasting an Obama win in that state's Jan. 8 primary. Clinton won by 2 points.

"It is relatively easy to predict general elections, as very little of the electorate is up for grabs -- maybe 15 percent. But in a primary, voters generally like all the major candidates and will end up voting for the nominee, regardless of who wins. In New Hampshire, Clinton, (former North Carolina Sen. John) Edwards and Obama all had very high favorability ratings prior to the election, meaning that voters were not that wedded to any of them."

In Ohio, 77 percent of Democrats proclaim themselves satisfied with their party's presidential candidates, the Dispatch Poll indicates. And Edwards is still drawing 3 percent of the vote.

Smith, who used to run the Ohio Poll, said that two days before New Hampshire's primary 21 percent of Democrats still had no idea who would get their vote.

The Dispatch Poll shows that 16 percent of the 2,308 Democratic respondents to the mail survey, conducted Feb. 21 through Friday, were at least "somewhat likely" to change their mind.

Eric Rademacher, University of Cincinnati political scientist who now conducts the Ohio Poll, says surveys in the Buckeye State involve at least as much risk as those in other states. He noted that the Ohio ballot still lists candidates who have dropped out of the race, and he wondered whether someone popular with Ohio Democrats such as Edwards could affect the outcome by pulling a "meaningful" percentage of the vote.

"In an election that basically comes down to two weeks of actual on-the-ground campaigning, the timing of polls may be critical," Rademacher said. "Estimating the actual composition of turnout, both in terms of geography and in terms of ratio of Democratic to Republican turnout, also will be important."

The biggest challenge for pollsters is making sure they are sampling the all-important "likely voter." The Ohio Poll asks a series of questions to determine respondents' interest in the upcoming election and their intent to vote; those responses are plugged into a scale rating their probability of voting.

Quinnipiac University in Connecticut has been conducting polls in Ohio for a few years, but its exact method of determining likely voters is regarded as so important that it is kept secret. In general, respondents are asked a series of questions on their interest not only in the election but in politics in general, as well as their voting history, a spokeswoman said.

The Dispatch Poll's historical accuracy has relied on a simple correlation: Mail poll ballots are distributed randomly to thousands of registered voters across the state; those who fill out and return the ballot usually are highly representative of actual voters. The poll only goes to registered Democrats and Republicans for a primary election.

Dispatch reporters Catherine Candisky, Alan Johnson, James Nash, Jonathan Riskind and Jim Siegel contributed to this story.

drowland@dispatch.com