The 1980 census, a billion-dollar endeavor that will employ 265,000 headcounters and the most sophisticated demographic techniques know to man or computer, will end up with the wrong total.

The Census Bureau already knows that it will miss some people - probably a few million - in the decennial census next spring. The bureau's demographers hope to be better that the 1970 census, when an estimated 5.3 million Americans were not counted, but they know the undercount problem cannot be eliminated.

The inevitability of error has sparked a spirited dispute about how the census should determine its final state-by-state population numbers. Should the final count be adjusted upward to account for the people missed? Or should the 1980 figures be restricted, as in every previous census, to the number of people actually counted?

If the undercount were evenly distributed among geographic regions and population sectors, this debate would probably be limited to scholarly monographs in demographic journals. In fact, though, the undercount question has sparked an intense political controversy.

The politicians are involved because of a basic fact of demographic life: undercounts hurt Democrats more than Republicans.

The undercount problem is most severe, the Census Bureau says, in minority neighborhoods in the cities, where people are suspicious of government and less willing to cooperate when federal enumerators come around asking questions. Since these neighborhoods tend to be overwhelmingly Democratic, Democrats say the undercount undermines their political base.

"Everybody knows that Democrats are harder to count than Republicans," says Rep. William Lehman (D-Fla.), who has been a leader of Democratic efforts in Congress to include an undercount adjustment when the 1980 census is tallied.

An undercount can hurt a region in two key ways.

Since the apportionment of the 435 seats in the House of Representatives is based on population, a state where the census misses people may end up with fewer House seats than its population would warrant.

Preliminary estimates indicate that New York state, for example, will lose at least three of its 39 House seats after the 1980 census, all from urban Democratic areas. Some New Yorkers claim the loss would be no more than two seats if all the residents of New York City ghettos were counted.

Census data are also used in the formulas that allocate about $50 billion annually in federal aid. An undercount can deprive an area of its legitimate share of Washington's largesse.

In addition to these regional concerns, national ethnic organizations complain that the undercounts impair the political clout of particular minority groups by failing to show their true strength.

Everyone - demographers, Democrats, and Republicans - agree that the undercount is a problem that must be dealt with. But how?

The Census Bureau's answer has been to concentrate on improving its ability to find people in the first place. It has undertaken a public relations campaign to convince people to cooperate, and it has promised to enlist minority group enumerators for minority neighborhoods.

Congressman Lehman, thinks that is not enough. "Your're talking about people who don't trust the government. They're just not going to cooperate."

What Lehman and many minorities would like to see is an adjustment to the totals achieved by the actual head count. "It is a sampling technique," Lehman says, "You extrapolate from your estimates of how many people were missed, and add that in before you report the final number."

But Republicans see in that proposal a potential for political tinkering. "I have no objection to counting every minority group member," says Rep. Jim Leach (R-Iowa), "but let's not start counting political phantoms."

Leach is worried that "adulterated statistics" could mean that "eastern urban centers could all too easily be given additional congressional representation and funding assistance at the expense of rural states."

The Census Bureau says it is still undecided about the adjustment quandary, but demographers there say they they do not yet have adequate information on which to base an adjustment.

"We have a problem to develop the necessary methodologies," says the bureau's Morris Gorinson. "There is a good chance we could develop valid adjustment factors at the state level but it doesn't seem likely that we could do it accurately for cities or counties."

After Congress stalemated on the adjustment dilemma last year, it asked the National Academy of Sciences to decide whether the census numbers could be accurately adjusted. The academy's panel ended up with a compromise answer - the numbers should be adjusted for federal aid purposes, but not for apportioning House seats - that satisfield nobody.

At the moment, the question is still undecided. The Census Bureau is assembling a panel of experts to take another look at the undercount problem. Congress is gearing up for another round of debate.

"You just can't help thinking," says Republican Leach, "that this whole question is going to get a political answer in the end."