1. The complete analysis may be found in Ira S. Lowry, Joseph S. De Salvo, and Barbara M. Woodfill,Rental Housing in New York City. II:The Demand for Shelter, R-649-NYC (The New York City Rand Institute, June, 1971).
2. Housing examples include R. N. S. Harris, G. S. Tolley, and C. Harrell, “The Residence Site Choice,”Review of Economics and Statistics, L, No. 2 (May, 1968), pp. 241–247; J. F. Kain and J. M. Quigley, “Measuring the Value of Housing Quality,”Journal of the American Statistical Association, LIV, No. 330 (June, 1970), pp. 532–548.
3. See U.S. Bureau of the Census, “New York City Housing and Vacancy Survey, Sample Design and Estimation Procedure,” public release dated August 20, 1965; reproduced in Paul L. Niebanck,Rent Control and the Rental Housing Market in New York City, 1968, Department of Rent and Housing Maintenance (January, 1970). When we refer subsequently to weighting, it will be understood we are using the census weights, supplied for sample households.
4. Strictly speaking, we should present standard errors offorecast, but, for our large sample sizes, this quantity, which is not a constant but increases for values of the independent variables farther from their means, is approximately equal to the standard error of estimate.
5. These households lived in controlled housing for which they paid cash rent. We chose only those records for which household income, household size, the age, sex, and ethnic background of the head of the household, and the household's gross rent expenditures were reported. We included cases in which contract rent was reported but utility payments were not; as explained above, we estimated utility payments for these cases. The records selected represented 708,000 households, about 57 per cent of all households paying cash rent for controlled-housing units. In nearly all cases, the records rejected lacked only data on income or age of head, items on which response was poor.