Missouri Census Data Center Program

Demographic Profile 3 Trend Report, 1990 - 2000: dp3_2kt

Usage Notes


These reports are a companion to the Missouri Census Data Center's
Census 2000 Profile reports. Those reports are an attempt to distill the most frequently accessed data items (variables) from the 2000 decennial census. These trend reports deal with a subset of the variables included in the dp3_2k reports, but combine these with comparable figures from the 1990 census. ("Comparable" needs to be in quotes for some variables.) The 1990 data items are derived from the Bureau's 1990 Summary Tape File 3 data tabulation. As such, these data are based on information collected on the census long form questionnaire. A similar trend report based on Summary File 1 (the "dp1_2kt" reports) have also created by the MCDC. However, there is very little data in those reports that is not also available here. These reports are much longer than their dp1_2kt counterparts, reflecting the much greater breadth of information reported on SF3 versus SF1 (long form versus short form).


Be sure to Read The Fantastic Metadata. We have featured it prominently in these reports through the extensive use of hyperlinks. The best of these are to the complete pdf "Variables" documents -- one for the
1990 variables and one for the 2000. These links appear prominently at the top of the report, just to the right of the link to the usage notes (this page). Also, within the body of the report, each table (in the table header line) has a pair of links (1990 and 2000) to the html versions of the same "Variables" metadata reports (same content, different format); these links take you directly to the location in the report pertaining to that table.

Finally, we have the "Notes" report (so called because the links to it appear in the report column labeled "Notes") that deals entirely with issues of compatibility across the two censuses.

Report Headers

These reports contain two header lines. The first is constant - "MCDC Demographic Profile 3 Trend Report, 1990-2000"; the second identifies the geographic area being summarized. It begins with the name of the area (if applicable) and then the standard geographic code, if any, used to identify the area. If available, a FIPS code will be used. Dashes are inserted between portions of these geocodes. For example, the geocode for a place will consist of the 2-digit FIPS state code, a dash, and then the 5-digit FIPS place code.

The blue-background table header lines identify the contents of the data columns. Follwing the Subject there are three data column pairs:

  • the values from the 2000 census
  • corresponding values from the 1990 census
  • and measures of the change in the values over the decade (the value for 2000 minus the value for 1990).
The Notes column is used to indicate which variables have footnotes that provide explanations regarding compatibility of the data item across the decades. Clicking on the notes icon ( ) for a data item will take you to the location within the footnotes document pertaining to that item.
An important thing to notice is that the value of the Pct column under Change is usually the change in the percentages rather than the percent change in the numbers. For example, when the Subject is "Urban Population" we have a Pct column for both 2000 and 1990 where we report the percentage of the population that was classified as Urban for each decade. The value in the "Change - Pct." column is simply the difference between these two percentages. It is the "Change in the percentage of population that is Urban"; it is notpercent change in Urban population (which would be calculated as the change as a percentage of the 1990 urban population.)

An important exception to this general rule is that if the item being reported does not have a Pct. value associated with it (for things like means and medians, for example) then the value shown in the Change-Pct. column is the percent change. For example if the item "Persons Per Sq Mile" has a value of 200 in 2000 and 160 in 1990 then we see a value of 25.0 in the Change-Pct. column (the percent increase in population density over the decade.)
If the data item (the Number column value) is already a percentage measure, then Change-Pct. will be empty. (Because we think a percent change in a percent would be confusing at best.)


General Layout of Tables

We have organized the data into 29 subcategories which we refer to as "tables". These are the same tables used in the dp3_2k (Demographic Profile 3) reports. Each table begins with a gray-background header row with the table's title, the Universe of the table and a pair of links to Metadata (1990 and 2000). The metadata links take you to specified locations (anchors) within the two html documents describing the two "filetypes" upon which these reports are based. These hyperlinks take you to detailed descriptions of the variables displayed in the tables, including a Definition column that shows how the variable was derived from the orginal SF3 tables. The metadata will also show a "Universe Variable" for each report variable. This indicates the variable used as the denominator for calculating the corresponding Pct variable. Note that there are also links at the top of the report to these same documents in pdf format (much nicer for printed copies).
These hyperlinks will display the metadata in a new browser window. If you click on more than one of the hyperlinks, it will reuse the previously opened window to display the new document. The display will not automatically be displayed over the current browser window. (So you click and apparently nothing happens; actually, it did, but in a window that is not currently visible on your screen.) To avoid this you can always close the windows that display the metadata before opening another; you can also right click on these links and specify that you want them to open in a new window.

The detail lines of the table are divided into 8 columns as described above. The Subject column identifies the data displayed on the line. Indentation is used to indicate hierarchies within the data. For example, in Table 24, "Owner Occupied Units" is indented within "Occupied Housing Units". Typically, the first detail line in a table contains the total universe count and is displayed in bold. All items that count or measure the entire table universe are left justified and bolded.

The Number column contains the count or other measure (mean, median or occasionally a percentage). The Percent column will be filled in if and only if the item is a count. It is usually (but not always) a percentage of the next higher level count. (As explained above, the variable used as the denominator for calculing the Pct column is shown as the "Universe" column in the metadata reports.)


Geographic Entities Over Time

Just as there are issues of comparability regarding specific data items (white population in 1990 is not the same in 2000; Median Household Income needs to be put in comparable inflation-adjusted dollars, etc.) there are also issues related to the geographic area(s) being summarized. While it is straightforward for most larger areas (the U.S., states and most counties), it can be very problematic for some levels of geography. For example, ZIP codes are infamous for changing their spatial definitions. The area comprising ZIP 63303 in 1990 may well be quite different from the 2000 version. It is important to know what, if anything, may have been done to "adjust" the geographies to make them comparable over time. It varies with the kind of geographic entity. Here is how we handle the different types:
  • Political entities (states, counties, places): These are reported using the definitions as of the census. If a city has doubled in size over the decade via annexation, the 1990 data will not be a summary of the current city limits, but rather the city as it was defined at the time of the 1990 census (January 1, 1990 to be more precise.)
  • ZIP codes: These are again based on the definitions used at the time of each census. For 1990 we use the data as published on Summary Tape File 3B. You need to be extremely careful when drawing conclusions about a trend within a ZIP code because it is very possible that the geographic entity summarized in 1990 is not the same as the one summarized in 2000. A good way to see if this is a problem is to look at the changes in Total Population and Persons Per Square Mile in Table 1. If the land area did not change over the decade then the Percent Change in these two items should be the same. But if population went up 20% and persons per square mile only went up 5%, then you know that land area must have also gone up, and hence the geographic areas are not the same for the two decades. Note: land area in square miles is not explicitly shown in the reports, but it is a variable on the extract datasets (LandSQMI), and it can be derived by dividing the total population by the persons per square mile.
  • Census Tracts and Block Groups: We only do trends at these levels within Missouri. The 1990 data used in these reports has been retabulated to estimate data values corresponding to the 2000 census codes. Thus you get trend data for census tracts that did not exist in 1990. You will notice that when you extract the 1990 data for these reports that the datasets accessed will have a name ending with "00", such as "motrs00.sas7bdat". This is an indication of a set that has been retabulated (the term "normalized" is sometimes used to describe this kind of restructuring) to 2000 geographic codes.
  • School Districts and State Legislative Districts: Only available for Missouri. The school district data has been retabulated to the 2000 district definitions. Likewise, the state legistlative district data has been retabulated to the current (i.e. as redrawn for 2002) legislative districts.
  • Metro Areas: These are as the time of each census. So the areas will not always correspond to the same geography.


Printing the Tables

These tables area readily printable in portrait mode for most setups. It takes 7 pages to display the report using IE default settings. In IE (but not in Netscape at the moment) the column headers will repeat on each printed page. The name of the area being summarized is included in the html document title, which will normally print at the top of each page.

Downloading the Data

At the bottom of the report you will see a row of hyerlinks to important related pages. The second of these is "Extract Data via Uexplore/xtract". This refers to the MCDC's
Uexplore web application for accessing our data collection. There are separate links for the 1990 and 2000 data. These links bypass the usual explore-the-directory phase of accessing a dataset, and take you directly to the xtract (data extraction utility application) first screen with the appropriate dataset already selected. The page includes a link to the online tutorial for the uexplore application (you may have to scroll to see the link -- located just below the line
First time users ....
You may need to specify a filter to avoid getting data for the entire dataset (this would typically involve coding a filter based on SumLev and/or State variables.) You will also have the opportunity to select variables so you do not have to take the entire 400 or so variables (it varies slightly with geographic level and year). In fact, if you are using a csv file to load into Excel, you must limit the extraction to 256 variables or less. There is also a 65,xxx row limit in Excel, so you may have to code filters and break the data into subsets.

Displaying Specific Tables

Used mostly be the developers as a handy testing option, the dp3_2kt application supports a tables parameter. It allows telling the program which specific tables you want to have displayed. For example, specifying the URL with parms
would result in a display of just tables 7, 8 and 9 for Adams county, Illinois. If the parms is not specified then the default is to print all 29 of them.


Linking to the Tables

The Missouri Census Data Center welcomes other State Data Centers to create web pages that point to these reports by coding URLs that point to the reports. These are of the form:

where [PARMS} specifies what geographic entity is to be displayed. This is done using a series of mostly-2-character geographic parameter names and their values. Best way to see how it works is to use scan the menus and see the underlying URLs. Here are some typical examples:
PARMS coded Area displayed
&st=oh State of Ohio
&st=12 State of Florida
&st=ny&co=005 Bronx county, NY
&st=55&pl=53000 Milwaukee, WI
zi=65201 ZIP Code 65201
st=us U.S. Totals

Related Applications

For most geographic levels there will also be a hyperlink at the bottom of the report labeled "Related Applications". This link will take you to our intermediate "applinks" application that will allow you to choose from a menu of other reports/applications available for the geographic area. For example, you will be able to link to the corresponding "dp3_2k" report (SF3-based profile) and a comparable report based on 1990 census data (for many, but not all geographic areas nationwide). A few of these applications will only be available for Missouri and/or Missouri, Illinois and Kansas.

More Detailed Data

There is a lot more data available from the 1990 and 2000 decenial censuses, and even from Summary File 3 for those two decades. It is way beyond the scope of this document to go into details -- a good place to start would be at the Census Bureau's
Census 2000 Gateway page. This will lead you to American Fact Finder, where you can get just about everything that is available. For more data from the Missouri Census Data Center you might want to check out the Readme file in our sf32000 data directory. If you are familiar with all the tables available on Summary File 3 (if not, see the Table Outlines chapter of the complete SF3 Technical Documentation.

The full 1990 stf3 data is stored in direcory/filetype stf903. At the moment, our metadata for this filetype needs work.

More Detailed SF3 Tables for 2000

Once you have found just what table(s) you are interested in you can rather easily invoke our sf3tabgen utility application -- the one that we link to when you click on one of the "SF3 Table" links in the dp3_2k application. We also have a separate sf3tabgen page that can be used to submit such requests.

More Detailed SF3 Tables for 1990

In addition to the considerable power of American Fact Finder for displaying 1990 STF3 tables, there is still the old reliable 1990 Census Lookup application, one of the first and best web applications related to census data.

Go to the dp3_2k application
main menu.
Address questions and comments to Glenn Rice.