Boston — The old Eaton Hotel in Wichita, Kan., now contains condominiums. Up in Camden, Maine, contractor Jeffrey Weymouth has rebuilt a white-clapboarded duplex on Mount Battie Street into a four-unit rental property.
And here in Boston, a new public-private partnership is raising $15 million to rehabilitate 500 units of abandoned housing.
What do these examples have in common? All of them add residential units to America's housing stock.
But none of them counts in the most widely used statistical measurement of housing growth - the US Census Bureau's tally of housing starts.
''If it doesn't start off with a new piece of ground,'' says Harvard Prof. William C. Apgar Jr., ''it doesn't count for the census figures.''
Professor Apgar, who heads a new Housing Futures Research Program at the Joint Center for Urban Studies of MIT and Harvard University, frowns on the use of such indicators to predict housing trends - because, he says, they produce some grossly inaccurate forecasts.
By way of example, he points to earlier forecasts of new construction during the 1980s. Many of them, foreseeing substantial increases over the '70s, were ''incredibly optimistic,'' he says.
Instead, his data show that ''the 1980s are the start of a long-term, substantial decline'' in new construction and mobile-home placements.
Other housing specialists contacted generally support his conclusions. Selwyn Enzer, associate director of the Center for Futures Research at the University of Southern California, says the ''boom'' of the '70s will not continue. Pointing to underlying changes in population and the rate of household formation , he says, ''I think we're running out of people to drive the system.''
''Housing growth reached its all-time peak in the 1970s,'' says Anthony Downs , senior fellow of The Brookings Institution in Washington and author of a forthcoming book on rental housing. He says Apgar's conclusions are ''probably correct - unless we have low inflation and low (federal budget) deficits.''
These analysts all foresee higher real costs of capital and increasingly expensive housing construction, pushing more people into multifamily units. ''This will make an enormous impact on the kind of house the American will buy, '' says Professor Enzer.
Apgar's study, which takes issue with some of the accepted methods used to make housing predictions, suggests:
* The prominence given to housing starts as a barometer of the health of the nation's housing supply is misleading. His research shows that in addition to 20 .9 million units of new housing and mobile-home placements added to the nation's stock in the 1970s, some 4 million ''non-new-construction'' units also came into the market. ''That's a significant component of change,'' says Apgar - and one that he feels will increase.
* The Census Bureau has overstated the rate of ''household formation'' - the number of individuals or families setting up their own homes for the first time - in the crucial under-35 age group. Government projections (which have yet to be updated with data from the 1980 census) call for an increase of 4.5 million such households over the decade. Apgar's study scales this back drastically, to about 1.1 million. Part of his reasoning: the post-World War II baby boom has run its course. He also notes that household formation is slowing as couples marry later - or as individuals spend more years in university dormitories or in the armed services, or pool resources to set up households.
* There is no ''giant pent-up demand'' for housing ready to burst forth when interest rates drop. The flaws in previous predictions, he says, have ''nothing to do with the financial climate'' of the nation, but are due instead to ''misanalysis'' of the demographic, economic, social, and political factors driving the market.
* Those factors were insufficiently examined in major earlier forecasts of housing in the 1980s - by the National Association of Homebuilders in 1975, the US Forest Service in 1977, Data Resources Inc. in 1978, and the Urban Institute in 1980. These studies, he says, were ''basically tack-ons of the broad economic forecasts'' that paid little attention to local detail.
''There is no 'national' housing market,'' he says, explaining that ''you can't understand the national market without understanding the way that a whole series of local housing markets function.''
So while his model forecasts zero growth in the number of households for the mid-Atlantic states (New York, New Jersey, and Pennsylvania) in the 1980s - compared with 49.6 percent growth in the 16 Southern states - there will be pockets of growth in the North and of stagnation in the South, he says.
So far, Apgar's team is making no policy recommendations for public officials. He notes, however, that the housing-construction industry in the '80s will have a significant amount of unused capacity. He says late in the decade might be ''a good time to turn on the subsidized housing market'' - to get ''a couple of housing units built cheap by a housing builder who didn't have anything better to do.''