Ed,
Working with any one block group almost anywhere in the county (Manhattan
comes to mind as a possible exception) will have such high MOEs that the
data are not worth anything, must less your time! There just are not
enough interviews in a single block group of, on the average, 300 housing
units, even over five years, to produce decent data. The only point of
making BG data available is for aggregation, especially in the radius
analyses so popular in the private sector.
Patty Becker
At 08:59 AM 3/16/2012, you wrote:
Mara--Let me know if you hear
anything. I think this is a very perplexing question and worthy of
some research under the banner of "how to work with ACS data, tips
and tricks".
Attached is something I did that might give you "a little"
insight. On the attached slide I had a block group table with 5
different modes to work, the Total Population and Total Workers.
What I wanted to do was to figure out the total number of commuters along
with the MOE. Remember Census does not give us the Total Commuters with
the block group data. Doing the math to get Total Commuters was easy,
just add up the modes. However, the MOE became something of a guessing
game. I first used the formula for for calculating the MOE for more
than two estimates. I got one number 214 (with 4 variables) or 245
(with 5 variables) depending if you used the MOE associated with a cell
that had zero observations. Dealing with zero cells in ACS brings its own
issues and questions.
I then realized that the total commuters was actually the difference
between the Total Workers and those who Worked at Home, only two
variables. That allowed me to use a different MOE formula (which by
the way is the same even though I was subtracting values instead of
adding) and I got 209.
My point is that although small with only 4 variables I was introducing
some additional error by increasing the number of variables.
Exactly how much I wasn't sure. If I did not consider the zero cell MOE
influence, the added error was in the neighborhood of 2.3%. So what
I found is that as you increase variables your MOE calculation would then
have an additional MOE. Hey a new statistic, the MOE of the
MOE.
What I could have done is a few more Block Groups and found some with no
zero values so that I could see what 5 variables as opposed to 4 did but
that is where I think this is worthy of research by someone with a
statical mind.
mkaminowitz@baltometro.org wrote:
Hello,
I need to get a regional estimate and MOE composed of 6 counties. I know
the rule of thumb is that aggregating MOE's doesn't work well for more
than 4 units. Does anyone know is using 6 would cause the MOE to be
inaccurate?
*Mara Kaminowitz
*GIS Analyst
.........................................................................
*Baltimore Metropolitan Council*
Offices at McHenry Row
1500 Whetstone Way
Suite 300
Baltimore, MD 21230
410-732-0500 ext. 1030
mkaminowitz@baltometro.org
<
mailto:mkaminowitz@baltometro.org>
www.baltometro.org
<
http://www.baltometro.org/>__
------------------------------------------------------------------------
*Confidentiality Statement*
This message may contain legally privileged and confidential information
that is intended only for the use of the recipient(s) named above. If you
are not an intended recipient, taking any action based on the contents of
this message is strictly prohibited. Please immediately notify the sender
if you have received this message in error.
------------------------------------------------------------------------
_______________________________________________
ctpp-news mailing list
ctpp-news@ryoko.chrispy.net
http://ryoko.chrispy.net/mailman/listinfo/ctpp-news
--
Ed Christopher
708-283-3534 (V)
708-574-8131 (cell)
FHWA RC-TST-PLN
4749 Lincoln Mall Drive, Suite 600
Matteson, IL 60443
_______________________________________________
ctpp-news mailing list
ctpp-news@ryoko.chrispy.net
http://ryoko.chrispy.net/mailman/listinfo/ctpp-news
-------------------------------------------------------------------------------------------------
Patricia C. (Patty) Becker
248/354-6520
APB Associates/SEMCC FAX
248/354-6645
28300 Franklin
Road
Home 248/355-2428
Southfield, MI
48034
pbecker@umich.edu