What's New in the 2K Edition of

The MABLE/Geocorr Geographic Correspondence Engine

John Blodgett

Overview

The biggest thing that is new with this edition of the MABLE/Geocorr application is that it references the revised MABLE database, mable2k. This collection of geographic codes was built (almost entirely in the initial release) from the 2000 Census Summary File 1 geographic headers files. As such, it reflects the Census geography (tracts, block groups and blocks) as used for the 2000 decennial census and also all the related geographic codes that were relevant at the time (January 1, 2000). In addition, we have extracted several important geographic codes that became available with the "UA" version of the Bureau's TIGER/Line files. We were able to add the codes Urban/Rural, Urbanized Area/Urban Cluster, and PUMA5, the geographic codes used in the 2000 Public Use MicroSample 5 percent sample files.
Unfortunately, we were not able to retain all the older geocodes from the previous edition of MABLE, which as based primarily on comparable geographic header files from the 1990 Census. That database was created using 1990 Census Blocks as the fundamental geographic unit; the new database is, of course, based on 2000 Census Blocks.

New Geographic Content

The enhancement to the geographic content is not just replacing 1990 codes with their 2000 equivalents. There are also altogether new codes, which were not available in the previous MABLE database because the Census Bureau did not support them as standard geography. Probably the most important of these are the 3 School District codes (Unified, elementary and secondary.)

New in Version 1.1 are 108th Congressional Districts (previously these were a Missouri-only option but now they are there for the entire country) and the new CBSA codes(Core Based Statitistical Areas, aka Metropolitan and Micropolitan Statistical Areas as defined by OMB in June, 2003).

Other New Features

While the geographic data used is the most important difference in this version, there are also some important differences in the way the application does what it does. The user can expect to see the results of their queries in a fraction of the time it took in the previous edition. They are also going to see a somewhat different format for delivering the results. The report-format output can now be in pdf or html format, rather than just the plain text format of earlier editions.

A new option allows the user to specify a custom title to appear on their output report. This should be especially useful in conjunction with the new output format options, allowing users to generate reports that are not just quick-and-dirty working documents, but somewhat attractive reports suitable for posting to the web.

We had hoped to dramatically overhaul the user interface, but in the end decided to focus our efforts - at least for now - with just getting a working version available as soon as possible. So the changes to the form are pretty minimal and still reflect the level of html expertise I had in 1996. The up side to this is that for those who are already used to it, there is not much new to learn. We moved the Geographic Filter section up to be third instead of fifth on the page because we found it was used a lot more than the other two sections. We have also added a 4th geography type that can be used to filter the query: Urbanized Areas/Urban Clusters. So, if are just interested in a list of tracts, ZCTA's, or counties for a specific Urban Cluster you can enter its code and geocorr will ignore all geography that is not within the UC area.
Another new wrinkle has been added to the geographic filtering section. For the place, metro area and urbanized/urban selection lists you can now enter the value "-9999", which will tell the program you just want it to ignore any geography that is outside any such area. So, for example, entering a value of -9999 for the metro area filter codes would result in output that would exclude non-metropolitan areas.

One thing that has been modified rather extensively is the way we deliver the results back to the browser. In the previous versions you saw nothing in your browser until the application was totally finished, at which point you got a menu page which you used to select various pieces of the output, including a "summary log". The new version is more user friendly and pipes information about your request back to the browser as it is working on it. When you request multiple states, for example, it sends a message back to the browser each time it starts work on a new state. In general, we think users will appreciate a much more streamlined dialog with easier access to the results.

No New Help

At this time we have not created any new Help pages, though we plan to. In the mean time, users should be able to get most of the help they need from the Help pages from the previous version. While some of the details of what is described there have changed, the gist of what the program does and how the user directs it it to do it have not changed much.

Response Time

Users should notice a dramatic improvement in response time. The machine we run this application on now is rated at eight times faster than the one on which we ran the old geocorr. We did a national query relating places to counties in just over 3 minutes. This would have taken at least a half hour on the old machine. The response time is also somewhat faster because of an improvement in the way we restructured the mable database.

(Techies Only): New Interface Tools

While most users will neither understand nor care about this (feel free to go to the next paragraph), we have changed the way we invoke the application on our server. We are still doing all the work with a SAS(r) program, and MABLE is still a collection of SAS data files; but we are now using the SAS/IntrNet interface product in place of the home-grown perl-based interface of the earlier versions. The biggest advantage of this change is in the ease of debugging for the developer, which is totally irrelevant to the user. It also made it much easier to improve the interface, so that there is now a lot of as-we-go information being piped back to the browser.

Mirrors, Hardware and a New State Bias

For now, this application is not being mirrored. It runs only at this site, on a Unix machine (IBM RS-6000 running AIX) at the University of Missouri Columbia, being jointly supported by the Office of Social and Economic Data Analysis at MU (part of University Outreach and Extension) and the Missouri Census Data Center. In a shameless display of parochial favoritism we have modified the State select list so that Missouri appears first and is the default selection. We have gotten rid of the option to allow the application to randomly select a state, as the fun of that had worn off somewhat.

New Geocodes

Different Kinds of New

There are two kinds of newness associated with the geography: I suppose one could cite a third category of codes that have really been somewhat redesigned and/or renamed. This is a category of one, but a very important one: ZIP codes and ZCTAs. The difference is really not as dramatic as the name change might suggest. The way the Bureau created their ZIP geocodes in 1990 was not all that dramatically different than the way they created ZCTAs in 2000. But there is a difference. See our ZIP Resources Page for a discussion of the difference and a link to the Bureau's definitive web page on the subject.

Legislative Districts

We have included the 106th Congressional districts (i.e. the districts used in the fall, 1998 elections to elect the 106th congress, which was seated in January, 1999 and was the current congress at the time of the 2000 Census. The current (spring, 2002) congess is the 107th and I believe it is identical to the 106th, but you should double-check this if it is important. Based on the results of the 2000 census, redistricting and reapportionment took place and the results will be seen at the national level in the upcoming fall, 2002 elections when the 108th congress will be elected. We have included 108th Congressional District codes only for Missouri.

Somewhat similarly, the database contains the state legislative districts (Upper and Lower chambers, known in Missouri and many other states as the Senate and House, respectively) as they existed at the time of the Census. These districts have been redrawn, but we only have the "redrawn in 2001" (and effective in 2002) districts for Missouri. We'll probably be able to add these later, if and when the Bureau adds the new district boundaries to the TIGER/Line files.

School Districts

These are new. The Bureau adopted these codes as being official recognized and stored in TIGER. Unlike in 1990, census blocks are no longer split by school districts. (If a proposed block was determined to be in more than one district back in 1999 when the Geography Division was preparing the gegoraphy for the 2k Census, then the block was split into multiple new blocks (just as would be done if it was discovered that the block was split by a city boundary.) There are 3 kinds of districts, and most states/areas do not have all 3 levels defined. In Missouri, for example, there are no Secondary districts and only a small number of Elementary ones. Codes of all 9s are used to indicate no such district defined.

Urbanized Area/Urban Cluster

Also known as UAs and UCs. We already mentioned these above, as having been derived from the TIGER UA files. The Urbanized Areas are an old concept, though the criteria for defining them have changed and now we have the new codes for 2000 based on the new criteria and the new data. Urban Clusters are entirely new and should be one of the more popular codes. Note that a given census block can not be in both a UA and a UC. It can be in one or the other or neither, but not both. Note that under the new definition of Rural, an area is considered rural if and only if it is in neither a UA or a UC.

Miscellaneous Geographic Categories

We are experimenting with some new codes based on feedback from users. We had people who wanted to analyze geographic areas by things other than simple geographic codes. They wanted to see what portion of an area was included within incorporated areas, or within places or counties of a certain size category. We have created the following new geo categories:

Census Codes for Place and County Subdivisions

We have had numerous occasions when it would have been convenient to have a file that related FIPS codes to census codes for these geographic units. So we now have added these two new code options. The census codes were derived from the 1990 Geographic Information Coding Scheme (GICS) files for these 2 levels. As such, they will not be defined for any entity that was not in existence for the 1990 census (which may be useful information in itself.) It also means that these codes are an exception to the general rule that all geocodes are based on the 2000 Census SF1 geoheaders files (excluding the Missouri-only codes, none of which are based on SF1.)

Census Block Code Changes

If you haven't run into the change yet, let us warn you. The Bureau has changed the format and numbering scheme for census blocks for 2000. In 1980, they were 3-digit numeric codes and in 1990 they were 3-digit numeric codes followed by an optional 1-alpha-character suffix. In 2000, they are 4-digit all-numeric codes. The 1st digit is still the key to the block group (i.e. all blocks beginning with "2" within a census tract constitute block group 2 in that tract.)

Another trivia item related to census geography is that the Bureau has dropped the "/BNA" terminology for the equivalent of census tracts in areas where there were no local tract committees to define census tracts. They are now referred to simply as "census tracts".

Missouri Only Geocodes

We have provided a number of extra codes for Missouri, because we only have them for Missouri. These are almost all county-based codes, such as various planning districts. The only sub-county codes being used thus far are the newly drawn legislative districts as mentioned above.
The Beale, LMA and Commuting Zone codes are a carry-over from 1990 and have not been updated. We hope to update these and provide them for the whole country eventually.
Address questions and comments to the author.

Return to MABLE/Geocorr2k.