For those SAS-Lrs who are interested in the Census 2000 data, I have
made available more detailed data on my site. All data models,
meta data, Zip, County tables and SAS matching code are based
on the Census data model. You can tie column names,
geography and table names directly to the exhaustive Census
published data and documentation.
This should make quality control with other sources of Census
data rather easy, as long as other sources can tie to Census meta
data.
I am approaching the Census data from Zip+4 geography, I have tried to
select the closest Zip+4, based on latitude and longitude, to each of
the
8,262,363 census blocks(entire US, DC Puerto Rico). I can then match
the block data to internal customer lists based on zip+4. If a match
fails on the 9 digit Zip+4, I drop last digit and use zip+3 (
summarized record ), if that fails I drop another digit ...
The SAS matching software implements the drop last digit logic. You
can use the Zip level ethnic data ( 43,019 rows and 288 columns) and
the SAS software to concatenate ethnic data to your Zip level lists.
If anyone out there has a 100Gb of ftp space, I am willing to build
and
store the SAS data model and make the SAS tables available to all.
This model makes the raw Census 2000 counts easy to use.
I only ask that the site follow the free software/data GNU model.
Roger DeAngelis
49 Spackenkill Rd
Poughkeepsie NY 12603
845-463-2770
http://members.aol.com/xlr82sas/utl.html