Questioning of the role of polls in elections has begun to rival criticism of the news media. The public may be reacting to what it sees as a concentration of power in the hands of a relatively few news organizations, particularly the magazine-newspaper-TV media combines that have added extensive polling programs to their news and analysis operations. This is a ''liberal'' reaction to a potential abuse of communications power. Added is a populist annoyance toward what is viewed as a communications elite.
On more practical grounds, some analysts claim - and others disclaim - that early projections of likely winners, based on exit polls in sample districts, before balloting ends can skew the voting patterns of those who have not voted. Although there was vastly more polling throughout the current election than in 1980, however, voter turnout was up fractionally over the previous election. For the presidential race, the TV networks again reported early in the evening, before Western balloting had ended, that the Reagan landslide was in the making. But surely anyone who had observed the pattern of survey results over any period - the last week or back to January - could have had little doubt about the outcome.
In fact, the structure of this election was evident four years ago. What is most striking about comparing this election and 1980 is their similarity. Jimmy Carter in 1980 and Walter Mondale in 1984 both won 41 percent of the vote. Carter won 45 percent of the women's vote; Mondale, 42 percent. Carter took 42 percent of the Roman Catholic vote; Mondale, 44 percent. The Democrats' decline among young voters was evident in 1980. And Reagan captured 26 percent of the Democratic vote both years. There are some important differences: Reagan picked up 11 percent more of the white vote this year than in 1980. But Jewish voters swung back to the Democratic column, 66 percent for Mondale.
How do we know? From exit polling, of course. In this case, from the CBS/New York Times election day surveys for the two elections. Without these data, we would have to wait for the definitive University of Michigan post-election survey, which turns out volumes of data tapes for academics - usually reported 10 months after the election at the American Political Science Association annual convention. Or for census bureau reports.
What would we do in the meantime? Without such data on how people voted, and the companion responses about why they voted that way, the public would have to leave it to pundits and politicians to say what it all means. Why not, as now, make public such data - which politicians pay pollsters to provide for them privately - and let the public interpret it for themselves?
It would be a pity if competitive reporting on election night, and overly aggressive interviewing, were taken as excuses to regulate or suppress this modern adjunct to reporting and research.