E-4 Population Estimates for California Cities and Counties, January 1, 1981 to January 1, 1990

This report provides the most current population statistics for California cities and counties and for the State, from the 1981 through January 1, 1990.

Population estimates from 1-1-81 through 1-1-90 , have been extracted from an E8090CITY report which is a historical data series consistent with both the 1980 and 1990 censuses. The 1-1-90 population estimates are from the current data series based on the 1990 Census. All population estimates contained in this report are consistent with the 1990 census.

Methods: Population estimates are initially prepared for each city and the unincorporated portion of each county using the Housing Unit Method. These estimates are summed for each county and adjusted so they equal county controls. The county controls are prepared using two separate estimating methods: Ratio-Correlation (regression) and a Composite Migration Method using drivers’ license address changes. These county estimates are then controlled to a State population total.

The 1990 Census population is the benchmark for all of the estimates. The county populations are estimated by adding the respective average population change, as measured by the methods to the previous year’s figure, and then adjusted to a State control. A brief description of the Housing Unit and county methods follows.

Rounding: These estimates are rounded according to population size as displayed below:

Population Size Rounded to Nearest
0-99 5
100-4,999 10
5,000-9,999 25
10,000-49,999 50
50,000-14,999,999 100
15,000,000 and over 1,000

The 1990 decennial census population was not rounded as census figures represent actual counts.

City Estimates

Housing Unit Method. This method is used to estimate total housing units, occupied housing units, persons per household, and non-households — the “group quarters” population. Data from the 1990 Census establishes a benchmark for each of these elements. Housing units are estimated by adding new construction minus demolitions to the census benchmark of housing units; adjustments for annexations are included. Occupied housing units are estimated by adding the change in residential electric customers to the benchmark of households. The independently calculated housing units and occupied housing units are then compared to obtain a vacancy rate and to evaluate their reliability.

A critical element in the estimation of population by the Housing Unit Method is the persons per household. This variable is calculated directly from census data and then adjusted for subsequent changes evaluated on the basis of the following indicators: Current Population Survey, county level trends, intercensal trends, special census findings in cities with similar characteristics, housing types constructed, and shifts in the number of resident school children attending grades 1-8 per household.

Population in group quarters is the final data element used in the Housing Unit Method. The summed total change over the year for all reported facilities are added to the census benchmark of group quarters population.

After the estimates are completed the household populations are adjusted to the county controls. The controlling process is applied because the county population estimates are more accurate than the Housing Unit Method population estimates. This accuracy is attained through the use of certain estimating procedures requiring data which are only available at the county level.

County Estimates

The county population estimates were developed by adding the average change in the results of the following independent methods to the latest census population.

The Driver License Address Change Composite Migration Estimating Method (DLAC) is a method in which migration of the population under 18 years of age is estimated using change in school enrollments, and migration of the population 18 to 64 years of age is estimated using address change on the California drivers’ license file. The number of migrants is estimated from address changes by using the 1990 Census relationship of population to driver licenses for ages 18 to 64. The resulting estimates of migration are added to the survived cohort of the household population under 65 years of age. Also added to this number are estimates of immigration from abroad including undocumented aliens, group quarters population, and an estimate of the population age 65 and over based on Medicare statistics.

The Ratio-Correlation Method uses a multiple regression equation based on selected symptomatic indicators (registered vehicles, school enrollment, births and voter registration) to estimate the change in a county’s proportionate share of the State’s population between the 1990 Census and the estimate date. The equation used to prepare the estimate is the following:

Y = 0.0507 + 0.0492(A) + 0.3711(B) + 0.4851(C) + 0.0515(D) where Y is the population variable and A, B, C, and D represent changes in the distribution of births, school enrollment, auto registration, and voter registration. Group quarters population are added to the calculated household population.

Accuracy

The estimates and changes shown here are subject to estimating error. Variation from actual population trends are inherent in estimating procedures because the correlation between the data series used are all affected by factors other than population change. The methods used to develop the estimates have been tested and modified through comparison with the results of censuses. The estimate for the State population as of April 1, 1990 was -0.29 percent lower compared to the April 1, 1990 Census. The mean absolute difference of the county estimates produced for April 1, 1990 compared to April 1, 1990 Census was 2.90 percent. For cities the difference was 5.75 percent.

NOTE: These estimates are provisional and subject to change.