The experts said it was close, but the voters -- stubbornly irreverent as usual -- made it a landslide. What happened?

To be fair to both the voters and the experts, a lot happened this year that was unique -- a televised debate one week before Election Day, a last-minute international drama, and more. The voters had good reasons to be uncertain to the last, and the experts had better reasons than some of them admitted to be confused.

What happened this fall vividly demonstrates the limitations of polling a a means of predicting election outcomes. The experience of 1980 is a reminder that pollsters can never predict exactly what distribution of Americans will vote on Election Day, so the polls can never be precisely right. Nor can the pollsters overcome the mathematical facts of life that govern their trade, so they cannot eliminate the substantial, inevitable margins of error even if they conduct perfect polls, which they seldom if ever do.

This election also reaffirms the voters' right to inflict surprises on all the experts, a right they haven't exercised so dramatically since 1948. "Dewey Deemed Sure Winner Today," the headline in this newspaper trumpeted on Election Day in '48. This year most of the pollsters said "too close to call." Both these mistakes reveal a good deal about the fallibility of pollsters and also the persistent unpredictability of the American electorate.

For the best-known public pollsters, 1980 was not a good year. None of them got close to the correct final result, though all of them did put Reagan ahead, and Louis Harris predicted flatly that Reagan would win -- by a margin of 5 percent.

However, the two candidates' private pollsters, who continued to survey voters through election eve, did foresee the final outcome accurately -- The Washington Post determined that fact last Tuesday, before the votes were counted. Carter's man, Pat Caddell of Cambridge Survey Research, predicted Monday night a 10-point Reagan victory, right on the money. Richard Wirthlin, Reagan's pollster, apparently foresaw an eight- or nine-point margin.

The fact that Caddell and Wirthlin did get it right will contribute to the myth that polls really can be accurate predictors. In fact, as the pollsters admit, when done with great care by meticulous, intelligent people, polls can produce results that probably come close to predicting an election outcome in some circumstances. "Polls are much better as explainers than as predicters," according to Barry Sussman, The Post's editor in charge of polling.

The basic statistical probabilities in polling are seldom explained in detail, but they are important. According to the laws of mathematics, a perfectly designed, perfectly executed poll of 1,500 people (a common size for a national survey) will reflect true national sentiment on that issue at that moment within a margin of error of plus or minus 3 percent 95 percent of the time.

That means, first of all, that when a pollster says a race is tied 50-50, he could just as accurately say that A led B by 53 to 47, or that B led A by 53 to 47. All three results, or any in between those extremes, are equally likely because of the poll's margin of error. Pollsters give precise figures because they think the press and public demand them. In an interview years ago, Lou Harris said it would be "much more honest" to give the full range of probability instead of one number. "But I've said this to the press before," Harris said, "and the answer back always is: 'You're chicken.'"

Second and equally important, because the probability of "success" for a poll like this can only be 95 percent, one poll in 20 or five polls in 100 will simply be wrong -- maybe a little wrong, maybe a lot. But neither the pollster nor the public will know which of the poll results they see are the wrong ones -- they look just like the "right" ones.

With these mathematical facts in mind, the performance of the pollsters in 1980 looks better than the way the pollsters themselves and the news media described their findings. If the final Gallup poll, for example, was accompanied by disclaimers like the above, the 47-to-44 margin for Reagan that Gallup predicted wouldn't look bad. Given the margin of error, Gallup's 47-to-44 result could have been 50 to 41, almost exactly right. Of course, if the margin of error was stretched in the other direction, that final Gallup Poll could have shown a 47-to-44 Carter lead as well.

As a matter of fact, the Gallup organization gave clear warnings to its subscribers in the textual material that accompanied its final poll. But this newspaper and most other Gallup subscribers dropped the warnings and just ran the 47-to-44 figure. (The Post did quote Gallup as saying its polls had never before uncovered "such volatility and uncertainty.")

Nevertheless, some of the polls were wrong and others inconsistent. Harris and Wirthlin, Reagan's pollster, saw a trend toward Reagan beginning even before the debate and continuing strongly afterward. Gallup saw a weaker trend starting after the debate. The Washington Post poll, sampling on Wednesday and Thursday of last week, did not pick up that trend, though it did see some movement to Reagan and extraordinary volatility -- that is, voters shifting back and forth between candidates. The CBS-New York Times poll saw no trend to Reagan in polling that ended Saturday. Caddell claims to have seen a strong trend to Reagan right after the debate, then a swing back to Carter that had put him back in the lead by Saturday. No other poll saw this movement, and the implication of Caddell's analysis -- that more than 4 million voters switched from Reagan to Carter to Reagan again in the course of a week -- is challenged by numerous other pollsters interviewed for this article.

Did the CBS-New York Times poll simply miss the trend to Reagan? "That's always possible," admitted Warren Mitofsky, the scholarly director of polling for CBS. Mitofsky personally believes that the debate, not the last weekend's Iranian events, moved the swing voters to Reagan, though his own poll did not show clear post-debate movement to the Republican.

The Post did two national polls and eight statewide surveys on Wednesday and Thursday of last week. One of the national polls -- the freshest sample -- gave Carter a four-point lead, the other gave Reagan the same margin. The state polls showed Reagan ahead marginally in all the big states except New York, where The Post had a clearly faulty poll giving Carter a 16-point lead. As explainers The Post polls were more revealing; by reinterviewing the same panels of voters nationwide and in the big states at an interval of several weeks. The Post polls showed that 30 percent of the voters had changed their minds during September and October, and that 10 million still were undecided less than a week before the election.

This last fact probably explains much of the difficulty the pollsters had this year. Whether voters began switching to Reagan Thursday or Saturday or even Sunday, the fact is that millions did switch in the final days. Polls take snapshots, not motion pictures, and sudden late shifts in opinion easily elude them. Those shifts also give the pollsters good excuses. "In our best judgment," George Gallup Jr. said in an interview after the election. "Reagan's lead grew steadily onward after Gallup stopped polling at 2 p.m. Saturday.

Another fact about this -- or any -- election that throws off the pollsters is the makeup of the real electorate. Mitofsky at CBS and Sussman at The Post, for example, expected an electorate this time that would be 52 or 53 percent men, according to extensive polling of voters leaving the polls on Tuesday. Since men voted much more heavily for Reagan than women, this change alone would have added a point or two to Reagan's strength in the final polls.

But, of course, the precise makeup of the electorate can never be predicted, especially when barely half of the adult population bothers to vote. So some polling error is inevitable, because pollsters must make assumptions about the electorate in order to "weight" the results of the surveys.

The Post's polls surveyed registered voters, but one-fourth of all registered voters didn't vote on Tuesday. Other polls sought to screen out unlikely voters, but as Lou Harris admitted in an interview, such efforts are never completely successful. This problem, too, guarantees some inaccuracy.

Finally, some voters lie to pollsters -- how many no one can say for sure. Mitofsky of CBS said his final preelection polls gave Reagan 5 percent of the black vote, but on Election Day 14 percent of black voters picked the Republican. The Voters

If the pollsters failed to explain in advance what the voters would do Tuesday, what does explain their behavior? It is probably premature to suggest a definitive answer, but a few facts and a hunch or two seem apt.

First of all, the Reagan landslide was unlike any other in modern times. The last two landslide winners, Lyndon B. Johnson in 1964 and Richard M. Nixon in 1972, won more than 60 percent of the popular vote. Reagan won 51 percent. Reagan piled up an electoral vote landslide that made Jimmy Carter look like Herbert Hoover, but he did so with the votes of just 26.7 percent of the voting-age population of the United States -- 51 percent of the 52.3 percent turnout.

Second, people really did change their minds a lot. The Post's resurvey of voters, suggesting that 30 percent changed their minds between September and the end of October, is a good indicator. Also high numbers of voters stayed undecided far longer than in other elections. Voters were genuinely unsure.

Prof. Samuel Popkin of the University of California at San Diego is a political scientist who worked during the election with Caddell's Cambridge Survey Research firm. He suggested a theory for the movement of voter opinion that might explain this unusual election.For most of the campaign, Popkin suggested, the swing voters saw it as "a choice between two futures," but at the end, it turned into "a referendum on the last four years," and when it did, Carter's candidacy collapsed.

In other words, swing voters could look at the election from two different points of view -- and many of them may have bounced back and forth between the two. One point of view was relatively serious; it involved weighing two distasteful alternatives for president, comparing the known shortcomings of Carter to the more problematical risks of Reagan. Polls all fall showed how voters were struggling with that comparison.

When the Carter campaign was working according to plan, the swing voters concentrated on the comparison, heard the negative things Carter was saying about Reagan, and worried about a Reagan presidency. When voters did that, Carter did well in the polls.

But there was a second general point of view that voters could take -- a more typically American skepticism about whether the incumbent deserved another chance. Harris asked respondents to his polls all fall if they agreed or disagreed with this statement: "There's just no way Jimmy Carter's first four years in office justify another four." In mid-October 49 percent in the Harris sample agreed, 46 disagreed. From then on, in polls taken every few days, the percentage agreeing kept rising. By Monday it was 54 percent.

Sometime during the last week of the campaign, if Popkin's analysis is apt, voters who had been weighing two evils suddenly decided to rely on their inveterate American optimism about the future, and also on their inveterate American competitiveness. There just was no way Carter's first four years in office could justify, for them, another four. So in the best American tradition, the voters decided to fire the coach. The exit polls suggest that there was more negative than positive voting on Tuesday.

In the televised debate, Reagan probably helped push viewers in the direction of a referendum on Carter with his brilliant summation. "Are you better off than you were four years ago?" Reagan asked. How many Americans could say yes?

The incomparable gifts of hindsight suggest that Carter helped encourage a referendum on his administration -- indeed, he may have ensured it. It was Carter who built up the image of Reagan as some sort of political ogre; when it came to the debate, all Reagan had to do was stand up and smile to shatter the entire Carter strategy. Then the inevitable anniversary of the seizure of hostages -- an event that was on the calendar all year, hardly a surprise -- came along just in time to remind the voters of Carter's most humiliating failure. The Carter campaign had made no evident preparations to deal with that devastating development. The flap over possible last-minute release of the hostages may have only aggravated popular discontent.

The exit polls also show that a lot of Americans who ordinarily would have voted -- most of them for a Democrat in a "normal" year -- just stayed home. In 1976 the electorate that voted was about 35 percent independent; this year only a fourth of those who voted called themselves independents. This left a more partisan real electorate in a year when about 90 percent of the Republicans voted for Reagan, but just 65 percent of the Democrats picked Carter. (This was another unexpected distortion in the real electorate that helped throw off the pollsters, too.) The fall-off in independent voters who are generally more likely to be ticket-splitters could help explain the unexpected success of Republican candidates for the Senate, too.

To begin with, at least, the new coach has the same disheartening material to work with that his predecessor failed to convert into a winning combination.Well, Vince Lombardi turned the Redskins around, didn't he?