The familiar Democratic coalition, born in the hard times of the New Deal era, re-emerged in yesterday's election.

Early this morning, NBC predicted that the Democrats would win 21 additional seats in the House -- a vigorous resurgence by any measure -- while CBS projected a 27-seat Democratic gain. Either outcome would produce Democratic majorities in the House comparable to any Franklin D. Roosevelt and Harry Truman enjoyed between 1939 and 1952.

Surveys of voters leaving the polls showed that union members, Roman Catholics, Jews, blacks, the poor and the unemployed were giving strong majorities to Democrats in a year in which economic issues -- summed up in the catch word "Reaganomics" -- overshadowed all other popular concerns.

But the traditional Republican constituencies -- white Protestants, higher-income voters and non-union families -- were sticking with their party, too.

The new "gender gap" phenomenon, in which women vote in greater numbers and more heavily Democratic than men, was not clear from yesterday's returns. Men appeared to be voting in somewhat greater numbers overall, but women were giving a bigger share of their vote to the Democrats. Nationwide, according to ABC, women gave 59 percent of their votes to Democrats, 39 percent to Republicans. By comparison, 55 percent of men voted Democratic and 42 percent, Republican.

These estimates and generalities are based on exit polls conducted by ABC and NBC. And, as is always the case, there were glaring exceptions to the apparent trends, best reflected in continued Republican control of the Senate.

In Indiana, for example, where unemployment exceeds the national average and has hit the automobile industry especially hard, the incumbent Republican senator, Richard G. Lugar, easily won reelection.

Next door in Ohio, where suffering is equally great, Democratic Sen. Howard M. Metzenbaum also won reelection as an unremitting foe of President Reagan's economic policies.

The shape of the Democratic vote was reflected in these findings from the networks' exit polls:

* Those who consider unemployment the most serious problem facing the nation were voting Democratic by more than 2 to 1. In those districts with the highest unemployment, Democrats enjoyed substantial leads.

* Blacks disapproved of Reagan's policies by more than 10 to 1, and in some areas were voting by about the same margin for the Democrats. Democrats were winning in four of the five Southern districts represented by Republicans that had been targeted by black groups. Of 23 black Democrats running for Congress, 16 were projected as winners and black Republican candidates were losing in 10 of 11 races. A notable loss came in Mississippi where a black Democrat, Robert Clark, failed to win a House seat.

* Union members were giving heavy majorities to Democratic candidates -- 69 to 29 percent.

* Jewish and Roman Catholic voters likewise showed a much clearer preference for the Democrats than they did in recent national elections.

* Protestant, upper-income and non-union voters, as usual, favored the Republicans.

Nationally, according to the exit polls, voters disapproved of Reagan's handling of his job by a ratio of about 52 to 48 percent. They disapproved of "Reaganomics" by nearly 6 to 4. And, by the same margin, they think Reagan should not seek reelection.

But there were crosscurrents and paradoxes in some of the findings. While giving Reagan bad marks on the economy, voters in the NBC sample were evenly split over whether his economic program had helped or hurt the country. They were evenly split over which party is best able to handle the economy, and by 6 to 4 they said their own financial fortunes would be best served if the government devoted more effort to controlling inflation than to the problem of unemployment.

One of the more interesting findings came from an Alliance Born AP/NBC exit poll: 6 percent said the Reagan economic program was a success; 37 percent said it was a failure; 48 percent said the program should be given more time to succeed.

The split in opinion was perhaps best illustrated by the indecision of independent voters. They voted Republican by 6 to 4 in 1980, but one network's exit poll yesterday showed them almost evenly divided between the Democrats and Republicans; another network's poll showed the independents giving Democrats a healthy majority.

One of the clearest indications of a rebirth of the old Democratic coalition was found in New Jersey, where Democrat Frank R. Lautenberg was elected to the Senate over Republican Millicent Fenwick. Data from the exit polls revealed that Lautenberg received 88 percent of the vote cast by blacks, two-thirds of the Jewish vote, 61 percent of the blue-collar vote and 58 percent of the union vote. He lost the rural vote, the white-collar vote, the upper-income vote and the Protestant vote.

In Connecticut, where Republican Sen. Lowell P. Weicker Jr. defeated Democrat Toby Moffett, the patterns were roughly the same, with Moffett winning majorities among Roman Catholics, union members, and blue-collar workers. But Weicker carried the suburbs and small towns, and won the majority of the votes cast by whites and non-union families.

There were clear regional dimensions to the traditional Democratic coalition, some of which seemed to reappear in yesterday's vote. Republicans made a good showing in New England and there was a resurgence of Democratic strength in the South.

The demographics and behavior of the American electorate -- who votes for whom and why -- have fascinated politicians, academicians, media consultants and special-interest groups for years.

In studying voters, academicians seek professional wisdom and doctoral dissertations; politicians seek to know where the votes are; consultants seek prescriptions for advertising and get-out-the-vote campaigns; television networks seek audiences through instant explanation and analyses; special-interest lobbies -- women, blacks, Hispanics, Moral Majoritarians, labor unions and political action committees--seek to demonstrate their power at the polls.

Rosalie Whelan, executive director of the National Women's Education Fund, put the matter in perspective: "The difficulty of measurement will always be that in victory, everyone wants to claim a leadership role. If, in fact, generally 'progressive' candidates supported by women's rights advocates along with women candidates win big on Nov. 2, the victories will naturally be claimed by their blue-collar, black, Hispanic, labor, environmentalist, nuclear freeze or other special-interest or demographic-group supporters. That's part of the political system."

There is another part of the system that many demographic analyses are unable to evaluate. That is the irrationality of voter behavior.

It is symbolized by a Chicago woman who, in 1960, told a University of Michigan research team that she had voted for John F. Kennedy after watching his television debate with Richard M. Nixon. She couldn't recall any issues raised in the debates but had concluded that some movement in Nixon's left eye raised doubts about his character.

Nevertheless, there are several well-established historical voting patterns. In most of the national elections over the past 50 years, the Republicans have been favored by white, Protestant, high-income, college-educated, business and professional people, by farmers, by white-collar workers and by people on the gray side of 40 years of age. The Democrats have been favored by manual laborers, Catholics, Jews, union members, blacks, the young, the poorest and the least educated. The 1982 returns reaffirmed those patterns.

And all through that period, Democrats have outnumbered Republicans, which accounts for the fact that the House of Representatives has been controlled by Democrats after every election but two -- 1946 and 1952 -- in the last half century.

What interests the demographers and political analysts are the shifts that occur in particular elections within these partisan groups. Relatively minor shifts can have major consequences.

This year, there is a high level of interest in one demographic group -- women. Until the 1960s, the turnout of women voters was far less than the turnout of men. They were, on the whole, less educated, less interested in politics, more deferential to the opinions of men. They were also slightly more Republican in their partisanship, according to studies by the University of Michigan's Survey Research Center.

By 1964, however, a majority of women had shifted political loyalties; they voted more heavily Democratic than men. And by 1980, they were turning out to vote in greater numbers than men. It is estimated that in 1980, women cast 5.5 million more votes than men.

Since women represent 53 percent of the voting age population, they are now the majority force in American politics in every state except Alaska, Hawaii, Nevada and Wyoming.

A few days ago, pollster Louis Harris reported: "Whatever gains the Democrats make . . . can be attributed almost exclusively to the women's vote. Women are concerned about their perceived inequality in society, especially in employment opportunities, pay rates and promotion possibilities. They sense that Republicans are insensitive and even hostile to those aspirations."

This partisanship by women transcends the view of some feminists that women should vote for women purely on the basis of gender. In this year's New Jersey Senate race, for example, a pre-election poll by the Bergen Record showed women more likely to vote for Lautenberg then for Fenwick. There were other setbacks for women candidates; 41 of 61 women running for the House appeared to be losing early this morning, along with four women candidates for other major offices.

Black voters represent another demographic group anxious to prove its clout in 1982. Black voters are overwhelmingly Democratic -- between 80 and 95 percent identify themselves that way -- and have been for years. Their political strength, however, has been limited by the low turnouts that are characteristic of all low-income groups. Their incentive to vote this year theoretically was strengthened by hard times. They suffer the highest unemployment rates and are, proportionately, far more dependent on the federal benefit programs that have been pared back by the Reagan administration. The results yesterday confirmed their Democratic allegiance in impressive fashion.

Other definable interest groups, as Rosalie Whalen noted, are anxious to claim credit and demonstrate their electoral power this year -- public employes, labor unions and political action groups of every description. Unfortunately, except in a few districts, their power will be difficult to measure on the basis of network exit polls and the like.

One special interest group that may have difficulty claiming success is the National Conservative Political Action Committee, which has sometimes inspired unnatural fears in the hearts of its enemies. NCPAC targeted 19 senators for defeat in 1982. Shortly before midnight, 11 of those senators were winning their races, three were leading and none was behind in the results that had been tabulated.

The federal employes' lobby lost a bit of political lustre, too. In Northern Virginia, where federal employes and retirees represent 40 percent of the electorate, two Republican congressmen appeared to have overcome the opposition of employe groups to win reelection.