IN THE WAKE of the landslide victory by Ronald Reagan and Republican candidates nationwide, red-faced election officials are once again attempting to explain away a continuing complaint about the American political system -- the fact that election results, as currently collected and counted, simply do not accurately reflect the polls taken in the final weeks of the campaign.

Though voting has become an accpted and important part of the American system, complaints continue about the scientific accuracy and usefulness of the results.

"I can prove to you beyond a shadow of a doubt that election results do not reflect a proper cross-section of the pollees," says Elmo Cavil, head of Altoona Survey Research, a widely known polling firm. "It disturbs me deeply that important political decisions continue to be made on the basis of this outmoded method of sampling public opinion."

Cavil cites his own poll results, which showed the presidential race as extremely close, with President Carter holding a slight advantage. However, when the vote totals were announced on Nov. 5, challenger Ronald Reagan had amassed a winning margin of more than 8 million votes.

"This result was well outside the margin of error," Cavil says. "I can only conclude that something is grievously wrong with the way the election was conducted."

Elected officials, for their part, admit they have a problems attracting a proper cross-section of the potential pollees -- whom they call, in pseudo-scientific jargon, "the electorate" -- to their voting places.

Bradley Headstone, chief of poll accountability analysis at the Federal Election Commission, points out that local election boards do use technical variables such as voting hours, poll locations, balloting methods and even pencil width to influence voters to turn out.

"Most of us in this business conscientiously try to get our results to reflect the polls," he said. "It's a problem we haven't licked yet."

One crude method of adjustment used in the past, Headstone says, was the "factoring in" of probable votes by potential voters who had died before the election date. "In recent years, this has gone out of fashion in most places," he says. "Perhaps we should study reviving the 'cemetery vote,' so to speak."

Headstone bravely contends that voting techniques have improved "tremendously" since the early days of elections. "In 1876, for example, the voting results at first appeared to show a landslide victory for Samuel W. Tilden," he says. "But after a thorough majority of pollees actually preferred the Republican, Rutherford B. Hayes. We haven't had that kind of embarrassment in many years."

Nevertheless there will continue to be surprises of the kind we saw on election day, Headstone admits. Longtime analysts liken this year's mistake, not to the oft-cited "Truman upset," but to the "Landon mistake" of 1936, when bungled ballots produced a "result" that was drastically out of line with a poll by Literary Digest, then one of the nation's leading news magazines.

But Headstone defends the practice of voting on the grounds that voters and politicians find the results informative and useful. And, he says, people just plain like voting.

"In some parts of the country, people have been voting for many years, and they have come to expect it," he says. "In some other places, like the South, it has only become really popular since the repeal of the poll tax. But even in those areas, our polling results show that people enjoy it, and that they would miss it if we dropped it from the campaign process altogether."

But Cavil and others continue to demand reforms. One suggestion Civil advances is that voters be required to cast their ballots by voice, over a mock telephone.

"This would as nearly as possible duplicate the actual polling situation," he points out. "I'm not saying it would end the abuses, but it might limit the damage somewhat."

Cavil notes that the United States once had this system -- known as viva voice or "live voice" voting -- and that "it worked rather well. Besides, it would expedite the work of network vote projectors tremendously."

The problem with the present system, Cavil says, is that "quite frankly, people lie to their ballots."