© 2000 University of Southern California.

Version 1.01

Release Date 7 July 2000

Ethington, Kooistra and DeYoung, Los Angeles County Union Census Tract Data Series, 1940-1990 was created with the support of a grant from the John Randolph Haynes and Dora Haynes Foundation. It is published and distributed by the University of Southern California to facilitate a wide range of research on the development and condition of Los Angeles County. Los Angeles County, the largest county in the United States, is of great interest to researchers from all disciplines of the human and natural sciences. Until now, no uniform census data were available to chart the development of this important metropolitan region through the period of its most important changes and critical decades. This data set is organized with the goal of providing maximum uniformity and comparability across all census years and census geographies. It includes aggregate data describing the characteristics of U.S. Census Tracts for race-ethnicity, occupation, education, housing, age and median household income, house value, and rent. It was assembled from two principal sources: 1) the print-published US Census tables for the years 1940, 1950 and 1960, and 2) the digitally distributed "Correspondence" data files created by the California State Department of Finance for the years 1970, 1980, and 1990. All data have been proportionally aggregated to fit the 1990 census tract geography. The boundary data are provided in ESRI's ArcView© shapefile format, and the numeric data are provided in the standard dBase© spreadsheet data file format (.dbf).

No fee is charged for the use of these data, and no fee may be charged for use or distribution. Publications and reports based on the data set must site it properly. The citation must include the following:

In addition, we request that users send us a copy of any publications, research reports, or educational material making use of the data or documentation. Printed matter should be sent to:

Philip J. Ethington

Department of History

University of Southern California

Los Angeles, CA 90089-0034

Electronic material or URLs for web sites should be sent to:

philipje@usc.edu

All available documentation for these data are included in this User's Guide document. No further support is available, but improved releases are planned, and users are invited send comments to the Principal Investigator, Philip J. Ethington, at philipje@usc.edu.

Ethington, Kooistra and DeYoung, Los Angeles County Union Census Tract Data Series, 1940-1990, Version 1.0, consists of a single table of 1,656 records (rows) and 252 variables (columns). The main data table, called la4090v101.dbf is in dBase IV © format and may be converted to any other format for quantitative analyses. The data set also includes the associated ArcView © shapefile to utilize these data in a Geographic Information System (GIS). These files must be kept together. They are called la4090v101.sbn, la4090v101.sbx, la4090v101.shp, and la4090v101.shx. ArcView © uses the la4090v101.dbf file as its "attribute table." Care should be taken when altering this file so that the shapefile continues to function with it. We recommend copying and renaming the main data file before beginning alterations.

Missing data are only certain as such if the cell is empty. Users must beware of vestigial sources of missing data, however, in for the form of zero values. These are most likely to occur in the "median value" variables. Data were taken directly from the sources listed below. If the source reported zero, we collected that value, even though a zero median home value might be highly unlikely. Users must use their own judgement as to when and whether to code these zeroes as missing. Users must also beware of the blank cells when converting these data to other formats, because some software does not support empty cells.

The data in la4090v101.dbf were prepared over the course of a four-year project, with the goal of assembling consistently comparable data from each of the six census years, 1940, 1950, 1960, 1970, 1980, and 1990. It is also intended that these data can be expanded backwards and forwards in time. Due to resource limitations, not all available tract-level data could be collected. The uniform method of collecting these data, however, make it possible to augment the data, so long as additions can be aggregated by census tracts. Further, not all variables are available in all census years. A core set of "union" variables, describing race-ethnicity, occupation, education, and median home rent and value information are available across all census years. In general, there are more variables describing occupations in the 1940-1960 years, and more variables describing median income, age, and marital characteristics in the 1970-1990 years.

The table below lists the variables by their generic variable names, and shows which are available for each year.

Sources:

The data in this data set were drawn from the following sources:

U.S. Bureau of the Census, Sixteenth Census of the United States: Population and Housing Statistics for Census Tracts. Los Angeles-Long Beach, California. (Wash, DC: GPO, 1942).

U.S. Bureau of the Census, U.S. Census of Population: 1950 Vol. III, Census Tract Statistics, Chapter 28 Los Angeles California and Adjacent Area Selected Population and Housing Characteristics (Wash, DC: US GPO, 1952).

U.S. Bureau of the Census, U.S. Censuses of Population and Housing: 1960 Census Tracts Final Report PH(1)082 Los Angeles-Long Beach, California. Standard Metropolitan Statistical Area (Wash, DC: US GPO, 1962).

State of California, Department of Finance 1970-1980-1990 Correspondence Data Series (Sacramento: California State Census Data Center, 915 L. Street, Lower level, Sacramento, California, 95814, 1996).

Methods:

The unique feature of the Ethington, Kooistra and DeYoung, Los Angeles County Union Census Tract Data Series, 1940-1990 is that all data can be studied according to a common geography for every year, from 1940 through 1990. In order to achieve this uniformity, all data were fitted to the geography of the 1990 census tracts. Census data or the years 1940, 1950, and 1960 were originally collected and stored in GIS shapefiles according to the census tract boundaries of those years. The 1970 1980 and 1990 data had already been fitted to the 1990 tracts by the California State Department of Finance (see specific methods below). We performed the fitting operation on the 1940, 1950, and 1960 data by imposing the 1990 tract boundaries on top of the boundaries for each of those years, and, using the "Union" function in ArcInfo ©, calculated a proportional distribution of data into the 1990 tracts. Obviously, certain assumptions and drawbacks are involved in this method. For the most part, census tracts in earlier years were geographically larger than those in following years. Thus, a 1940 tract with 1000 persons might have been split the Census Bureau into two or more tracts in 1950, to reflect growing population, again in 1960, and so on through 1990. Those original 1000 persons are assumed by our method to be spread uniformly across the space of the 1940 tract, and to land proportionally, by area, in the 1990 tracts that overlay that 1940 tract. Thus, assume that hypothetical 1940 Tract A was split over the years into five 1990 tracts, called Tract V, W, X, Y, and Z. Tract A was 100 square kilometers. 1990 Tract V lies completely within 1940 Tract A, and is 25 square kilometers. 1990 Tract W also lies completely within 1940 Tract A, and is 30 square kilometers. 1990 Tract X, Y, and Z lie both inside and outside of 1940 Tract A, however. Only 2% 1990 Tract X lies within 1940 Tract A (call that 2 square kilometers), 50% of Tract Y lies within 1940 Tract A (call that 30 square kilometers), and 80% of 1990 Tract Z lies within 1940 Tract A. The following person counts would have been assigned to the 1990 tracts: V=250; W=300; X=20; Y=300; Z=130. Note that 1990 Tracts X, Y, and Z will also pick up data assigned from the other 1940 tracts they overlap.

The 1970 and 1980 data were fitted to the 1990 census tracts by a different method. Below we have reproduced verbatim the State Department of Finance's description of its method:

" The 1970 – 80 – 90 database was created to simplify thematic mapping of historical California data for the years 1970, 1980 & 1990. The data for all three years are assigned to 1990 census tracts. Census geography has changed since 1970 making the process of relating 1970/1980 data to 1990 census tracts more difficult. The result of this effort, however, eliminates much of the difficulty and allows the user to easily compare census tract data between 1970 and 1990. The data, as assigned to 1990 census tracts, will accurately depict the demographic and economic changes for this period.

To help the user of this file better understand the relationship of 1990 census tracts and the 1970/1980 geography, a correspondence file is available that indicates how the 1970-1980 data were assigned to 1990 census tracts. If the user wanted to produce a tabular report of the data for 1970 or 1980, beware of the fact that the data cannot be considered OFFICIAL U.S. Census Bureau data or California Department of Finance data.

There are seven possible types of geographical changes between 1970 and 1990:

1. A census tract existed in 1970 and has not been subdivided:

1970	1980	1990
0001	0001	0001

The actual data appears for this tract in each year.

1. A census tract existed in 1970 and was subdivided in 1980:

   1970	1980	1990
   0001	0001.01	0001.01
   0001	0001.02	0001.02

2. A census tract existed in 1970 and was subdivided in 1980 and 1990:

   1970	1980	1990
   0001	0001.01	0001.11
   0001	0001.01	0001.12

   0001	0001.02	0001.21
   0001	0001.02	0001.22

   The actual data appears for these tracts in 1990 only. The 1980 data were assigned to the 1990 tracts based upon the population split of the 1990 tracts. The 1970 data was assigned to the four 1990 census tracts based upon the population split of the two subdivided 1980 tracts. The 1982 and 1970 medians are assigned without change.

3. A census tract existed in 1970 and was subdivided in 1990:

   1970	1980	1990
   0001	0001	0001.01
   0001	0001	0001.02
   0001	0001	0001.03

   The actual data appears for this tract in1990 only. The 1980 and 1970 data were assigned to the three 1990 tracts based upon the population split of the three 1990 tracts. The 1980 and 1970 medians are assigned without change.

4. A census tract existed in 1970 and was merged with another tract in 1990:

   1970	1980	1990
   0001.01	0001.01	0001
   0001.02	0001.02	0001

   In this situation the 1970 and 1980 data are aggregated and the medians are re-calculated.

5. Actual adjustments (other than subdivision) were made to a census tract boundary in 1980 and/or 1990. In these situations the boundary adjustments have been ignored. The assumption is that these changes are insignificant in respect to thematic mapping purposes and do not distract from interpreting change between 1970 and 1990 in a particular geographic area. Users must be cautioned that the boundaries in this file accurately describe 1990 census tract areas BUT may not accurately describe 1980 or 1970 census tract areas.
6. Census tracts did not exist in 1970 and/or 1980. In this situation Enumeration districts (Eds) or Census County Divisions (CCDs) were matched to 1990 tracts according to the following steps:

Below is a table with short descriptions of all variables. The "YY" in variable names stands for the census year: "40" (or 50, 60, etc) will appear in place of the YY in the variable containing the 1940 (or 1950, 1960, etc) data for that variable, if available.

This table is hyperlinked to the detailed data descriptions for each variable. Users are strongly recommended to read the extended descriptions carefully before analyzing these data.

(Note: Hyperlinks are not operable in this portion of the document)