Subject: Re: Bird counting
Date: Dec 4 17:03:21 1995
From: Burton Guttman - guttmanb at elwha.evergreen.edu



On Mon, 4 Dec 1995, Eugene Hunn wrote:

> Good point. My preference for CBC data would be to publish only one or
> two significant digits, e.g., 3.4 x 10e5. I doubt that the additional
> digits have any meaning whatsoever.

Let's say you can count small numbers of birds with reasonable accuracy,
so if someone reports, say, 15 birds or 25 birds you can accept the report
and figure that the error is small. Now suppose I say that the best you
can do with larger numbers of birds is plus or minus 10 percent. Suppose
someone reports a flock of 350 birds. Using the figure of 10 percent,
that means the actual number could be anywhere between 315 and 385, so
even the second digit, the "5," is suspect. A third digit would have no
meaning at all. I guess I'd accept the 5 as a best guess, meaning that
the number was more than 300 but less than 400, but I sure wouldn't have
much faith in it. Of course, I'm talking about flocks, seen all at
once. If you tallied up a few crows at a time and came up with 355 at
the end of the day, I'd believe the second digit and have a little faith
in the third.

Burt Guttman guttmanb at elwha.evergreen.edu
The Evergreen State College Voice: 360-866-6000, x. 6755
Olympia, WA 98505 FAX: 360-866-6794