I haven't looked at the files yet (MI just came out last week), but I'm
pretty sure what you're seeing is the results of rounding. This is a way
of introducing "noise" into the data in order to preserve
confidentiality. In my opinion, it's one of the best solutions (especially
considering the alternatives!), and helps users to note that the data are
really pretty noisy anyhow. In truth, there is no statistical significance
to the difference between, say, 7 and 9, in a single TAZ on any variable
that comes from the sample items on the census.
Patty Becker
At 12:52 PM 07/28/2003 -0400, you wrote:
All:
Has anyone else noticed that the values for counts in the CTPP Part 1 data
always seem to be rounded to multiples of 5 (except for 5 itself, which
usually seems to be 4)?
Also, it seems that the total for a variable rarely is the sum of the
other values for that variable. For example:
- If you add up the individual means of transportation, only
coincidentally will they add up to the total of all means of
transportation, using the Both Sexes fields in Table 1-002
- When you look at the same data in SF 3, the values are close but
different, clearly not rounded to multiples of 5 and definitely adding up
to the expected total (in this case, All Workers)
Peter
----------------------------------------------------------------
Peter Van Demark
Director of GIS Products and Training Phone: 617-527-4700
Caliper Corporation Fax: 617-527-5113
1172 Beacon Street E-mail: peter(a)caliper.com
Newton MA 02461-9926 Web site:
http://www.caliper.com
-------------------------------------------------------------------------------------------------
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(a)umich.edu