Unless I am missing something, Elaine's challenge has not yet been met or not responded to on this list. Anyway, while trying to IPF a table in R i ran across this resource (http://www.demog.berkeley.edu/~eddieh/datafitting.html) that might be useful to meet Elaine's challenge (there is a bunch of R code that will help meet the challenge).  Elaine has already given Age x Mode, and Gender x Mode. I am attaching Age x Gender for workers 16 years and older [no MOEs included and obtained after raking the marginals of tables B08101 and B08006] so that all the data elements are available to meet her challenge. 

Krishnan

On Fri, Mar 11, 2011 at 6:45 PM, <Elaine.Murakami@dot.gov> wrote:
Here is the challenge! I do not know how to do IPF, but I know that many
of you out there can do it!
Attached are:  ACS B08101  ; ACS B08006 for Maricopa County Arizona.
You can ignore the MOE cells for this IPF challenge.

Also attached is a 3-way tabulation using IPUMS for Maricopa County AZ.
All are using 2009 ACS.
You will note that carpool is not broken out, as the question is really
in 2 parts (what mode, and then how many people), so this table could be
re-done by creating a carpool variable, if you wanted.

We have an archived webinar on using IPUMS, if you are interested in
making your own tabs using ACS PUMS.
http://ctpp.transportation.org/Pages/webinardirectory.aspx

Elaine

-----Original Message-----
From: ctpp-news-bounces@chrispy.net
[mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Erlbaum, Nathan
(DOT)
Sent: Friday, March 11, 2011 11:42 AM
To: ctpp-news@chrispy.net
Subject: RE: [CTPP] 2010 Data Release Watch

Noted below is a question and a response that appeared on the list serve
earlier today and that suggests a solution method to creating data that
almost exists but is needed.  I am reposting this to the list serve,
because I believe that this might be a worthwhile effort for someone who
is adept and familiar with IPF to provide a valuable resources to the
CTPP community in much the same way the ACS MOE spreadsheets have
evolved to explain how to deal with this statistical concept in an easy
to use way for the practitioner.

Recently there was a post to the listserve by Jonnette Kreideweis about
the Census Conference in California and the call for papers.   There was
also a recent post about the Planning Methods Conference.

A spreadsheet that would answer the question posed is potentially a good
basis for a paper.  Further it helps share skills with the larger
practitioner community based upon solutions and techniques that others
have implemented.   When presented and discussed on a listserve then
many more people perhaps those who might never consider this as an
option may now have a solution to the very same problem or may move on
to adapt the method to other analysis areas.

I want to encourage anyone who has the time to give it a try, maybe a
student can use this to apply what they have learned, or a modeler can
cannibalize something he/she has already used, many may benefit from you
efforts.



----------------------------------------------------------
Nathan Erlbaum
Associate Transportation Analyst
Office of Policy, Planning & Performance New York State Department of
Transportation 50 Wolf Road, 6th Floor Albany, New York 12232

(Tel) 518.457.2967
(Fax) 518.457.4944
(E-mail) nerlbaum@dot.state.ny.us
(Web)  www.nysdot.gov




>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
Marcus Wigan wrote:
> ed
>
> a terribly simple question
>
> How best to do age group by gender by travel on the CTTP!
> best
> marc
>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

Ed Christopher's response:

Unfortunately, there is no clean way to get Age by Gender by Mode (Means
of travel), that I know about, directly out of the current 3-year CTPP
data product nor the Standard Census ACS products for that matter.
However, there are some options that will get you close but they will
take a little work.

Option 1.  In CTPP take table 12201 ( Age by Mode to work) and table
11203 (Age by Sex/gender)and run an IPF (Iterative Proportional
Fitting-fratar) routine between the two tables to get what you want.  At
least I think it will work.

Option 2.  Do the same with standard ACS product tables B08006 (Sex by
Mode) and B08101 (Mode by Age).  If you go this route I think these
tables may even exist in the 5-year ACS so you could do it at smaller
than Places over 20K population geography.

Option 3.  Just make your table using the PUMS data but you would be
limited to geographical areas of over 100K people.

Over the years we has suggested that people use an IPF routine to make
tables that might not exist in the various data products.  I would be
interested in hearing if anyone has actually done it.  Have you?



_______________________________________________
ctpp-news mailing list
ctpp-news@chrispy.net
http://www.chrispy.net/mailman/listinfo/ctpp-news




--
Krishnan Viswanathan
1101 High Meadow Dr
Tallahassee FL 32311