About the 1990 Census Geodatabases on WAGDA
|While the 1990 and 2000 census geodatabases constructed for WAGDA are similar in design, there are a few important differences that you need to be aware of in terms of how the databases are constructed, and in how the data itself has changed. Data between 1990 and 2000 is NOT directly comparable for a number of reasons. You can follow the general instructions provided for using the 2000 Census geodatabases, with the caveats for the 1990 version listed below. Visit the Census Bureau's webpage for the full documentation for STF1A, STF1B and STF3. Go to our geodatabase download page to get the data and metadata.|
- The 1990 STF1 dataset has fewer tables. Whereas the 2000 SF1 dataset has approx 40 tables, STF1 has only 13, labeled 0 to 9 and BX, BZ, and BXBZ. The PCT tables in the 2000 SF1 set that contain data at the census tract level were not tabulated in the 1990 census STF1 dataset.
- There is no dedicated geography table. The 2000 census geodatabases have a geography table that lists all of the identifier information for each record and unit of geography. In the 1990 geodatabases this information is not stored in a dedicated table. Instead, it has been lumped into the first data table, STF1A0.
- Field numbers have changed. Variables in the 1990 and 2000 census do not share the same field numbers; they are completely different. Sometimes there is only a slight difference: total population is in field P001001 in 2000 and in field P0010001 in 1990. In many instances, field numbers are completely different. For example, the population inside urban areas is in field P002002 in 2000 and in field P0040001 in 1990.
- The unique ID number is different. The unique ID field in the 2000 census for relating the tables to the boundary files is LOGRECNO. In 1990, the unique ID field is LOGRECNU. The numbers in these fields are not directly comparable, i.e. you cannot join tables from 1990 to tables in 2000 based on these fields.
- Block and Metro area data for 1990 is stored in a separate table. Unlike the 2000 geodatabases, where each data table contains census data for all summary levels, the block data for the 1990 census is stored in a separate table from the other summary levels. They are actually two distinct datasets that we have been combined into one geodatabase: STF1A, which has all of the 100% count variables down to the block group level, and STF1B, which has a small subset of 100% count variables down to the block level. All of the block data is stored in tables STF1BX, STF1BZ, and STF1BXBZ. A few other summary areas, notably metropolitan areas, are also only available in STF1B.
- Empty blocks are handled differently in the 1990 census. In the 2000 census geodatabases, each data table includes the data for all census blocks, even if those census blocks contain no population. In the 1990 census, blocks that have no population, such as blocks that were drawn to contain parks, forests, cemeteries, and facilities with no permanent residents, were excluded from the block data table, STF1BX. These empty blocks are stored in a special zero sum table, STF1BZ. This makes it easier to use the data for statistical analysis, but harder to map areas contiguously. In order to make the data more map-able, WAGDA has created a table called STF1BXBZ which contains all of the blocks in the data table and zero sum table. If you want to map data showing all blocks contiguously, including blocks with no population, use the STF1BXBZ table. If you want to completely omit areas with no population, use the STF1BX table. Unlike the 2000 census, blocks that contain only bodies of water were excluded entirely from all data tables.
- There are fewer variables at the block level. There were fewer variables tabulated at the block level for the 1990 census.
- There are some extra variables in the STF1 tables. The STF1 tables in the 1990 census include some variables that were not part of the SF1 dataset in 2000, such as the value of housing units. In 1990 this question was asked of the entire population on the short form, but in 2000 only a sample of the population answered this question on the long form, and thus it appears in the 2000 SF3 dataset and not in the 2000 SF1 dataset
- Some variables are tabulated differently. There are a number of instances were variables in the 1990 census were recorded differently from the 2000 census. For example, the 2000 census listed age by gender, so that there were two separate age tables for men and women. The 1990 census listed the age for all genders in one table. So, if you want see age broken down for the population as a whole in the 2000 census, you would have to combine the data from the two separate tables. On the other hand, if you want to break down the age groups by gender for the 1990 census, you would have more work to do. The 1990 STF1 dataset only has tables that break age down by gender and race: you would have to combine data from each race table to calculate total population by gender for each gender. This is but one of several examples of changes in tabulation.
- Some variables were created differently. Some of the questions from the 1990 census were changed in 2000. A notable example is the question for race. In 1990 respondents had to choose one racial category, but in 2000 had the option to choose a category that indicated that they had more than one racial background. This change has a profound affect on historical analysis, and must be taken into account when comparing 1990 and 2000 census data. For example, if you are looking at how the Asian American population has changed in a particular area over time, you must consider that some of this change is actual change caused by births, deaths, and migration, but some of the change may be due to the change in the question. Some persons who identified themselves as Asian Americans in 1990 may now have identified themselves as being from more than one race in 2000 since they now had the option to choose this category. This may give the false impression that the Asian population is growing slowly or declining in some areas, but in reality the change is due to people identifying themselves differently. This change affects all racial categories. Once again, this is but one of several possible examples of how a question has changed. Visit the Census Bureau's Difference in Questionnaires page for a list of all the changes between 1990 and 2000.
- Census geography changed between 1990 and 2000. Census geography is not static. If an area is experiencing population growth, census tracts, block groups, or blocks may be subdivided into smaller units to account for a higher density of people. If an area is experiencing population decline, then census geographies may be aggregated into larger units. In some cases, geographies may be redrawn with new boundaries all together, but this does not happen as frequently as subdivisions or aggregations. The Census Bureau's definitions of certain areas, such as metropolitan areas, also change from one census to the next. Visit the Census Bureau's Geographic Changes for Census 2000 Glossary for more information on general changes to census geography between 1990 and 2000.
- Not all census geography is available in 1990. For instance, the census bureau did not tabulate STF1 data at the Zip Code Tabulation Area (ZCTA) level in 1990.
- Only STF1 data is currently available on WAGDA. WAGDA is only providing the STF1A and STF1B datasets, which represent the 100% count data (short form), for the 1990 census at this time. We may assemble geodatabases for the STF3 sample data (long form) for 1990 at some future date.