Skip to main content
Log in

Accuracy of the housing unit method in preparing population estimates for cities

  • Published:
Demography

Resumen

Este artículo informa sobre los resultados de un test sobre la exactitud relativa del método de unidad de vivienda en la estimación de la población de ciudades. Fueron preparadas estimaciones para 47 ciudades de más de 50.000 habitantes en las cuales se han conducido censos eepeciales durante los años 1964–66. Esta sciudades fueron elegidas porque en general las ciudades que han afrontado los gastos de un censo especial son comunmente lugares que estan desarrollandose rápidamente o que han realizado incorporaciones de territorios vecinos a su ejido (anexiones); estimaciones sobre este tipo de areas estan sujetas a un margen de error mayor que las estimaciones sobre areas donde los cambios de población son más normales.

El test señala cinco caracteristicas del métoda de unidad de vivienda para la estimación de la población de las ciudades: (1) En general, el métoda da estimaciones que tienden a sobrestimar la población. Cuando datos sobre permisos de construcción fueron usados como base para las estimaciones, las desviaciones fueron positivas en 36 de lost 47 casos. En las ciudades donde las desviaciones fueron negativas aproximadamente la mitad había realizado importantes anexiones luego de 1960. (2) El uso de datos sobre los servicios públices en lugar de los permisos de construcción permite generalmente reducir el tamaño de los errores, a pesar de que las desviaciones continuan siendo más positivas que negativas. (3) Las desviaciones son menores cuando los valores de 1960 no son usados y el iamaño medio de la familia (hogar censal) es extrapolado. (4) Tanto con el uso de permisos de construccion como con el numero de usuarios de servicios públicos, el error promedio en la estimacion del numero de viviendas familiares (hogares censales) es alto. (5) La estimacion del número de hogares censales es un mayor contribuyente a los errores de estimacion de población que la estimación del tamaño medio de los hogares censales (para esie test extrapelaciones de los valores 1950–60 o asumiendo no cambios desde 1960). A pesar de que el alcance de esie test fué limitado, el metodo revelo una eficacia relativa aceptable, con errores promedio de 3.6 a 5.8 por ciente, con exclusión de las areas que estan experimentando grandes anexiones.

Summary

This paper reports the results of a test of the relative accuracy of the housing unit method in the estimation of the population of cities. Estimates were prepared for 47 cities in excess of 50,000 population in which special censuses were conducted during the years 1964–66. Because cities that undertake the expense of a special census are commonly areas that are experiencing rapid growth or large annexations, extimates for such areas are subject to wider margins of error than estimates for areas of more normal population-change patterns.

The test points up five features of the housing unit method for the estimation of the population of cities. (1) In general, the method yields estimates on the high side. When building-permit data were used as a basis for estimates, the deviations were positive in about 30 of the 47 cases. Of the cities in which the deviations were negative, about one-half had had substantial annexation after 1960. (2) The use of utility data instead of building-permit data generally reduces the size of errors, although here too there are substantially more positive than negative deviations. (3) Deviations are smaller when the average size of household is extrapolated than when the 1960 values are used. (4) When either building-permit or utility data were used, the average error in the estimate of the number of households was high. (5) The estimate of the number of households is a greater contributor to errors in estimates of population than is the estimate of the average size of household (jor this test, extrapolated from 1950–60 values or assuming no change since 1960).

Although the scope of the test was limited, the method made a relatively creditable showing, with average errors of 3.6 to 5.8 percent, excluding areas that are experiencing large annexations.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

References

  1. United States Bureau of the Census, “Inventory of State and Local Agencies Preparing Population Estimates: Survey of 1965,”Current Population Reports, Series P-25, No. 328, March, 1965.

  2. For California cities, Carl Frisen presented summaries of tests based on special censuses in the 1950 decade in the Proceedings of the Social Statistics Section, Annual Meeting of the American Statistical Association, (Chicago, Illinois, December 27–30, 1958), pp. 231–32. More recent tests have been conducted by Mr. Frisen's successors at the California Department of Finance, but these findings have not been published. (See also footnote 9.)

  3. A variation of this basic method used by the Bureau of the Census is discussed in some detail in United States Bureau of the Census,Current Population Reports, Series P-25, No. 371.

Download references

Author information

Authors and Affiliations

Authors

Additional information

The authors gratefully acknowledge the assistance of Mr. Jude J. Klein who did much of the research in the early stages of the study, and of Mrs. Sandra S. Miller who was responsible for compiling the basic data and for carrying through the computational phases.

Rights and permissions

Reprints and permissions

About this article

Cite this article

Starsinic, D.E., Zitter, M. Accuracy of the housing unit method in preparing population estimates for cities. Demography 5, 475–484 (1968). https://doi.org/10.1007/BF03208591

Download citation

  • Issue Date:

  • DOI: https://doi.org/10.1007/BF03208591

Keywords

Navigation