1 Feb
2011
1 Feb
'11
1:30 p.m.
Hello,
I realize the issues of data comparability (and--dare I say--data fusion) can be tough,
but I think Wade White is on the right track. I used the following analogy many years
ago, as part of a presentation at a transportation planning applications conference: if
you have one watch, you think you know what time it is. If you have two watches and they
don't show the same time, you become less certain as to what represents the truth.
Consider the following scenario: detailed journey-to-work flows are available from the
CTPP 2000, while county-to-county flows are available (at least for the larger counties)
from the 3-year ACS. If both datasets represent the best available depiction of the
"observed" in their respective years, does it make sense to use the latest
county-to-county numbers as "control totals" for adjusting the older TAZ-to-TAZ
(or tract-to-tract) flows to represent current detailed flows? Quite frankly it's not
crystal clear how to answer that kind of question. In many cases the updated TAZ-to-TAZ
flows might be just fine for most planning purposes, whereas in other cases such a
straight-forward factoring approach could give a false level of confidence as to the true
accuracy of the adjusted (synthesized) detailed flows.
Perhaps the best that can ever be done when examining less-than-perfect datasets is to
keep one's eyes open by first comparing the different datasets and then exploring
possible explanations for the differences: what happens next depends on the findings of
the exploration and what one wants to do with the data. Sounds like an area for some
useful applied research!
Ken Cervenka
FTA Office of Planning and Environment
Ken.Cervenka(a)dot.gov
________________________________
From: ctpp-news-bounces(a)chrispy.net on behalf of Wade L. White
Sent: Tue 2/1/2011 6:50 AM
To: ctpp-news(a)chrispy.net
Subject: [CTPP] RE: ctpp-news Digest, Vol 83, Issue 32
Penny:
I would chime in that my experience suggests all data and findings should be compared as
the differences, whether due to methodology, underlying assumptions, or actual observed
change can add great insight and "point the way" towards the answers desired.
Such comparisons often highlight differences and should help analysts start to comprehend
the weaknesses and strengths of data upon which they're building assumptions. It
doesn't make one set of data more "right" than another, just a different
perspective. I always ask "what does the comparison tell me?" Does the
comparison yield differences and similarities I expect? Are the findings consistent with
other "common wisdom"? If no, why not?
For example, take a few very different data sources: the CTPP, ACS, NHTS, QCEW, IRS,
D&B and InfoUSA. Each include very different ways of estimating employment. All are
incomplete but paint part of the picture. Taken together, I can see patterns and weigh
the relative strength of each one's methodology to meeting a specific need.
How are part-time jobs treated? How about sole proprietors? How about workers with more
than one job? Workers that work one day/week or month? 1099 workers? Volunteers?
Seasonal workers? Military? Are there things missing, discounted or even double-counted
in one vs. the other? Understanding Standard Error terms is one thing, understanding
utility and suitability is quite another.
Even trying to answer a simple question such as "how many jobs/workers are there at a
given moment in time" doesn't have a simple answer or at least there isn't
one dataset that will answer it depending on the definition of job/worker I'm looking
for to meet a specific need.
One "comparison" I was looking at in the 2000 vs. 2009 3 year data is a simple
"relative order" of destination county in one dataset vs. the other (row-wise
relative ranking). Comparing the 2009 3 Year data (off your wonderful website) vs. 2000
CTPP would seem to have merit. See below:
<https://webmail3.dot.gov/exchange/Ken.Cervenka/Drafts/RE:%20[CTPP]%20RE:%20ctpp-news%20Digest,%20Vol%2083,%20Issue%2032.EML/1_multipart/image003.png>
Are the changes consistent with what is known about the economic and development change in
the area? Observed traffic flows (yet more data) and changes in the flows over time?
What's missing or considered incompletely in each dataset? Do the differences in the
dataset suggest I need to use them to inform decision-making differently.
All fair and good comparisons (and questions) that help provide me some level of insight
both into data differences and change. For a "real-world" comparison of two
different dataset that has made a difference in understanding, think about the release and
subsequent re-release of the NHTS based on comparisons made between its findings and the
NTD for US transit ridership. The comparison highlighted additional considerations that
would need to be included in the weighting process for NHTS. Without the comparison, the
suitability of NHTS for certain transit analyses would be in question.
The more comparisons we make and questions we ask the better the data gets and the better
the understanding of its limitations. Of course, if we blindly use any data, caveat
emptor.
My 2 cents.
W
-----Original Message-----
From: ctpp-news-bounces(a)chrispy.net [mailto:ctpp-news-bounces@chrispy.net] On Behalf Of
ctpp-news-request(a)chrispy.net
Sent: Monday, January 31, 2011 11:06 AM
To: ctpp-news(a)chrispy.net
Subject: ctpp-news Digest, Vol 83, Issue 32
Send ctpp-news mailing list submissions to
ctpp-news(a)chrispy.net
To subscribe or unsubscribe via the World Wide Web, visit
http://www.chrispy.net/mailman/listinfo/ctpp-news
or, via email, send a message with subject or body 'help' to
ctpp-news-request(a)chrispy.net
You can reach the person managing the list at
ctpp-news-owner(a)chrispy.net
When replying, please edit your Subject line so it is more specific
than "Re: Contents of ctpp-news digest..."
Today's Topics:
1. RE: 2000 and 2006-08 work trip comparison question.
(Weinberger, Penelope)
2. RE: 2000 and 2006-08 work trip comparison question.
(Kendra Watkins)
----------------------------------------------------------------------
Message: 1
Date: Mon, 31 Jan 2011 10:41:28 -0500
From: "Weinberger, Penelope" <pweinberger(a)aashto.org>
Subject: RE: [CTPP] 2000 and 2006-08 work trip comparison question.
To: <ctpp-news(a)chrispy.net>
Message-ID:
<94A99461953E3341B89642B00D5C0B7D0705A4B4(a)AASHTO-MAIL.aashto.local>
Content-Type: text/plain; charset="us-ascii"
The CB does not recommend comparing an ACS based data set to a CB Long
Form based data set. The 2000 data represent a point in time estimate,
the ACS data represent a period estimate. Furthermore, the Census
Bureau recommends not comparing period estimates with overlapping years.
Penelope Weinberger
CTPP Program Manager
AASHTO
202-624-3556
http://ctpp.transportation.org/Pages/default.aspx
<http://ctpp.transportation.org/Pages/default.aspx>
It's just as bad to not make a plan as to blindly follow the one you
already have.
From: ctpp-news-bounces(a)chrispy.net
[mailto:ctpp-news-bounces@chrispy.net] On Behalf Of Seidensticker, Dan
Sent: Friday, January 28, 2011 4:18 PM
To: (ctpp-news(a)chrispy.net)
Subject: [CTPP] 2000 and 2006-08 work trip comparison question.
We downloaded the2006-2008 ACS county-to-county worker flow for Dane
County, Wisconsin from http://ctpp.transportation.org/Pages/3yrdas.aspx.
The question we now have...can that data be compared to the
county-to-county 2000 CTPP work trips to determine any statistically
significant increase/decrease? If so, how would one calculate the
margin of error?
Dan Seidensticker
GIS Specialist
Madison Area Transportation Planning Board:
A Metropolitan Planning Organization (MPO)
City of Madison Planning Unit
121 S. Pinckney Street, Suite 400
Madison, WI 53703
Voice: 608-266-9119
Fax: 608-261-9967
Email: dseidensticker(a)cityofmadison.com
www.MadisonAreaMPO.org <http://www.madisonareampo.org/>