We are hosting a short session to train users on using the Minnesota Population Center IPUMS . http://usa.ipums.org/usa/ The IPUMS project provides a user-friendly approach to using decennial Census and ACS Public User Microdata Samples (PUMS) including the newest 2005-2007 ACS file and the Census 2000, and many older decennial census files. These microdata samples are a portion of the full individual records and include all the variables. The data are NOT pre-tabulated.
If you need a quick tabulation that is not included in a CTPP or a standard Census file like Summary File 3, or the American FactFinder, then using the microdata may be a solution for you. The PUMS are restricted to larger geographic units (Public Use Microdata Area) to protect individual confidentiality.
For example, Nathan Erlbaum from New York State recently wanted to know the the impact of group quarters population on bicycle commuting. I ran a table using Census 2000 (includes group quarters) and 2005 ACS (excludes group quarters). Probably I should have compared 2005 ACS to 2006 ACS! By the way, the answer was that group quarters had a large effect on walking for the mode to work, and not much difference on bicycling.
You do not need to have statistical software installed on your PC. The SDA program in IPUMS can run simple tabulations, regressions, and logit models. We will learn basic techniques to select rows and columns, filtering (subsetting), re-grouping variable classifications (very important for continuous variables like travel time, departure time, age, income) and even try a simple regression. The on-line software cannot do as sophisticated variable processing as a system like SAS or S+, but it can provide analysts with a quick way to examine basic patterns that can be used to design more sophisticated analyses.
DATE: April 16, 2009 Thursday
TIME: 1 - 2:30 p.m. Eastern
Pre-registration is required:
http://www.nhi.fhwa.dot.gov/resources/webconference/web_conf_learner_reg.as…
I have reserved 65 slots. Hope to see you there!
IPUMS for Census Data
Main instructor: Katie Genadek, Minnesota Population Center, IPUMS project.
Elaine Murakami
FHWA Office of Planning
206-220-4460 (in Seattle)
I will be out of the office starting 03/17/2009 and will not return until
03/23/2009.
If you need immediate assistance please see one of the people acting for me
during my absence.
Kin Koerber - March 17- March 19
Melissa Chiu - March 20
For all CTPP-related matters please see Melissa Chiu.
I am not sure if others are pondering what the recent Federal Register
notice on ACS 5-year data products really means but I believe that there
are some “key” facts missing. Facts that would allow users’ the ability
to asses the true impact of the rules. For example:
Rule 7 of the Disclosure Avoidance rules states that “For the residence
and workplace tables where means of transportation (mode) is crossed
with one or more other variables, there must be at least three
unweighted workers in sample for each transportation mode in a given
place for the table to be released. Otherwise the data must be
collapsed or suppressed and complementary suppression must be applied.
There is no threshold on univariate tables.”
http://www.census.gov/acs/www/Downloads/proposal_acs5yearproducts.pdf
While this might sound very straight forward we know from the
“statistical collapsing” rules that were applied to the 3-year data that
only one collapsing attempt is made of the data and if the table does
not pass it is tossed out (suppressed). What we do not know about the
current rule is what will the collapsing hierarchy be, and will the
Census continue collapsing modes into each other until the threshold is
met? Both the Federal Register notice and the rules are silent on this
issue.
In some recent work done for FHWA by the Census Bureau we do know that
at a Tract level a great deal of data will be suppressed under Rule 7.
Using 5-year data from five of the ACS test counties, Multnomah, Lake,
Broward, San Francisco and Bronx we do know that if all 17 modes (means
of transportation) are used only 3 to 8 percent of the Tracts would have
enough data to be released. However, if we cut the number of modes down
to only 4 including Drove Alone, Carpooled, Transit, and other including
work at home somewhere between 5 and 37 percent of the tables will be
suppressed. And that is at the TRACT level! If per chance someone
wants to go one step up the modal ladder and split out “bike and walk”
and make work at home a separate category yielding 6 modal categories,
upwards of 80 percent of the TRACT level tables can expected to be
suppressed. At the Block Group the suppression will be only worse.
--
Ed Christopher
Resource Center Planning Team
Federal Highway Administration
19900 Governors Drive
Olympia Fields, Illinois 60461
708-283-3534 (V) 708-574-8131 (cell)
708-283-3501 (F)
Dear Everyone:
Today, the Census Bureau issued this Federal Register notice. The
5-year ACS is the first ACS product for small area geography, e.g.
census tracts and block groups. I have extracted KEY POINTS that will
GREATLY IMPACT data availability for transportation tables and result in
a lot of DATA SUPPRESSION at the tract and block group level. As this
FR was just issued today, we have not had a chance to estimate the
amount of suppression that would occur given these proposed conditions.
We are currently working on profile sheets using the 2005-2007 ACS
(3-year data) that was released in December 2008. We are finding many
tables to be suppressed. So, even with the population threshold of
20,000 for the 3-year ACS data, there is considerable suppression on the
current ACS standard table production, at least those tables of key
interest to transportation planners.
As most of you know, the first CTPP using ACS is a planned 3-year
accumulation from ACS 2006-2008. The table list will be re-submitted by
AASHTO to the Census Bureau in a few weeks.
The next CTPP product is envisioned as a 5-year accumulation from ACS
2006-2010. Because of the many rules the CB has established to protect
individual confidentiality not just for the ACS Standard tables, but
also for custom tabulations like the CTPP, we believe that the 5-year
CTPP for small area tabulation (e.g. TAZs) will need to rely on
synthetic data generated inside the CB using the ACS microdata records.
_________________________________________________________
*Federal Register Notice: The 5-Year ACS Data Products Release Plan
Today the U.S Census Bureau published its data release plan for the ACS
5-year data products in the Federal Register (E9-4803). Beginning in
late 2010, the Census Bureau plans to introduce 5-year data products
covering the January 2005 through December 2009 data collection period.
The release of the 5-year estimates will achieve the goal of the ACS to
provide small area data similar to the long-form sample data published
after Census 2000.
The Census Bureau is proposing to modify its current line of data
products to accommodate the 5-year estimates and is requesting comments
from current and potential users of ACS data products to help guide this
modification.
We invite you to review the 5-Year ACS Data Release Plan and provide
your response to the contact listed in the Federal Register notice.
Please follow this link to the Federal Register notice (PDF files)
posted in the Highlights section on the ACS Main page:
http://www.census.gov/acs/www/
Comments are due to the contact listed in the Federal Register notice by
April 20, 2009
>From the Census Bureau's pdf document:
2. Detailed tables with more than 100 cells cannot be released at the
block group level.
5. For the Selected Population Profiles, there must be at least 50
unweighted sample cases
over the 5-year period in the universe (specific population subgroup) in
a given
geographic area for the profile to be released.
6. For workplace tables, there must be at least 50 unweighted or 300
weighted workers in
sample over the 5-year period in a given workplace for the table to be
released.
7. For the residence and workplace tables where means of transportation
(mode) is crossed
with one or more other variables, there must be at least three
unweighted workers in
sample for each transportation mode in a given place for the table to be
released.
Otherwise the data must be collapsed or suppressed and complementary
suppression must
be applied. There is no threshold on univariate tables.
Elaine Murakami
FHWA Office of Planning
elaine.murakami(a)dot.gov
Note: please update my contact information. My previous email address
will likely expire on March 11.
206-220-4460
I am out of the office from Mon 3/2 - Fri 3/6 , returning possibly 3/9
weather permitting. If you have an IT issue, please carbon Bob Curtis
as he will proxy for me in my absence. If you have a GIS issue please
contact Aaron Westcott, I will have limited e-mail access but will be
monitoring my messages. If it is a technical question I'll get back to
you upon my return.