Messy polls blamed on volatile electorate Reasons varied, the analysts say

The answers: a) You can't trust any of them. b) You haven't decided yet. c) Go away, you don't want to talk.

The question: Why are public opinion surveys on the presidential race producing such seemingly erratic results, putting Arkansas Gov. Bill Clinton anywhere from 2 to 11 percentage points ahead of President Bush?


"It's a mess," said Glen Bolger, a Republican pollster, when asked yesterday to explain. "There's an unusual amount of volatility for this stage of a campaign."

Mr. Bolger and other analysts cited several major reasons for the apparently disparate findings, including Ross Perot's late re-entry into the White House contest; expectations that voter turnout this year will be higher than usual, but uncertainty about how to measure the size or impact of that growth; differences in methodology among pollsters; and a mix of voter disapproval of the status quo and discontent with the contenders being offered to change it.


Mostly, those interviewed stressed that many of the numbers being reported in the week before Election Day are not as out of whack with one another as they look.

For example, CNN polls reported Mr. Clinton leading Mr. Bush among likely voters by 2 and 3 percentage points, respectively, over the past two days, while many simultaneous canvasses put the margin at 7 points or more. But by adding the 3-point margin of error to CNN's surveys, and subtracting the same margin of error from the other polls, the results can be shown to easily conform.

Larry Hugick, the Gallup Organization pollster who conducts CNN's polls with USA Today, cautioned that voters should not put too much faith in any one survey -- particularly if it is a one-day "tracking" poll such as his own, which questions fewer people than do longer surveys. The smaller the sample, the greater the possibility that it will not be representative of the population.

"We have to accept some randomness in this business; these polls are not perfect," Mr. Hugick said.

Some other pollsters said results this year were especially difficult to analyze because many more people, up to half in some samplings, were refusing to participate.

To get a more accurate sense of events, analysts said, news media and voters should average results from all the polls they receive. Using that technique, Mr. Clinton appears to be ahead of Mr. Bush by 6 to 8 percentage points, about the margin both the GOP and Democratic campaigns are finding in sustained private polling.

While the Bush campaign seized on the CNN-USA Today surveys to argue that the presidential marathon had been transformed into a neck-and-neck race, other pollsters and Democratic aides asserted that faulty methodology by the Gallup Organization had led to an out-of-sync product.

"If you take the average of all the national polls," Mr. Clinton's lead is between 7 and 8 percentage points, said George Stephanopoulos, Clinton communications director.


The 2-point result came on the day Gallup switched from reporting its interviews with registered voters to including only those deemed most likely to cast ballots. Critics said the problem with making that change so late in a turbulent election season -- one in which voter interest seems to be especially high -- is that determining who will turn out is difficult.

Both politicians and pollsters said the most persuasive argument against paying too much attention to these polls was that they measure national sentiment, while the election is decided in the Electoral College.

That means state-by-state surveys provide a better gauge of what is likely to happen on Nov. 3. In every such poll for weeks, Mr. Clinton has shown enough support to exceed the 270 electoral votes required to win.