Three major newspapers were forced to make last-minute changes in the way they packaged their election coverage in today's editions because a computer problem prevented them from receiving national exit polling data.

The newspapers had planned to publish major stories that tracked voters' attitudes on a variety of issues based on interviews with voters leaving polling places. The interviews were conducted by a network news consortium, Voter Research and Surveys, formed to coordinate polling results and projections of winners and to end what some critics have called reckless network competition in projecting early election results.

National exit polls of voters' stands on House races and national issues were supposed to be available to subscribing newspapers beginning at 3:30 p.m., said Warren Mitofsky, executive director of Voter Research and Surveys. The data still were not available by 11 p.m.

Mitofsky said VRS had the data, but the computer program designed to weight them -- by taking into account regional variations -- did not work. Without such weighting, the results might overstate or understate nationwide trends.

"The survey came in properly, the data was all there," he said, but the computer program was not operating properly. "We have a problem in that program. The programmer has been trying to find it all evening."

"Everything else was working fine," Mitofsky added, including state-by-state exit polls -- except those in the District -- and winner projections.

The Washington Post, the New York Times and USA Today were the only newspapers to have ordered the national exit polling data, Mitofsky said.

Because they could not use the VRS results, there were last-minute turnabouts on deadline in newsrooms of the three papers.

"We have definitely had some problems here tonight," Jim Norman, polling coordinator for USA Today, said shortly before 11 p.m.

"How about building pages around data that wasn't available and having to change our game plan in the middle of the evening?" he said. "We've had to change the direction of our coverage. We had writers and we had editors and we had graphic artists, all who were depending on this information."

Mike Kagay, editor of news surveys for the New York Times, said, "While we were inconvenienced, it was not mortal."

VRS was established in March as a combination of resources and personnel from news divisions of CBS, ABC, NBC and CNN to collect and tabulate Election Day exit polls and winner projections in local, state and national elections.

The consortium's formation represented a truce of sorts among the networks, which had been criticized for releasing polling data so early in past elections that they influenced voters in states where polling places were still open.

Although Mitofsky said the national polling data were not as clean as he would have liked, networks used it to develop reports on voters' attitudes on such questions as what they considered to be the election's key issues. Mitofsky said the national data were not unreliable.

"It could be better but it's good information," he said. "The final step in the weighting may change the percentages a few points. Now I don't think on most issues that that's significant. I think a few points one way or the other is not going to change the prevailing opinion."

But newspapers intended to report national exit poll results in detail. Because the poll results were subject to change, it was too risky to use the detailed data, said Richard Morin, director of polling for The Washington Post.

"The simple fact of the matter is that the information in the national exit poll was not usable," Morin said.