Subject: Re: CBC Accuracy??
Date: Dec 7 13:35:33 1998
From: "Deb Beutler" - dbeutler at wsunix.wsu.edu


-----Original Message-----
From: Tom Foote <footet at elwha.evergreen.edu>
To: Don Baccus <dhogaza at pacifier.com>
Cc: tweeters <tweeters at u.washington.edu>
Date: Sunday, December 06, 1998 5:49 PM
Subject: Re: CBC Accuracy??


>> It's value lies in...

> 44,000 participants x $5/head :) :)

> Tom


Actually, that is a very important point. A professor at Kansas State
University, David Rintoul, has looked at the effects of the $5 a head price
the Audubon Society collects for every CBC participant and its effect on the
numbers of participants. What he found was not surprising: as the price of
participation has increased, the number of participants has decreased. Of
course, the natural assumption is people stopped participating as the price
increased. Kansas has a unique system; some of the CBC are published in the
official Audubon magazine but all of them are reported in the Bulletin of
the Kansas Ornithological Society. If you compare CBC that are published in
both locations, the number of participants is larger in the KOS Bulletin
than in the Audubon's official reports. Why? When David asked the
compilers of the counts in question, they admitted that they under reported
the number of participants to Audubon because you must have $5 for every
participant you report. Many people were unwilling to pay the money; rather
than discourage people from counting, the compilers just didn't report them
as a participant. Does this only occur in Kansas? Probably not.

This may not seem important but when it comes to actually using the data, it
can become a big problem. As others have noted, the number of people
counting and amount of time they spent counting (party hours) or the
distance they covered (party miles) is often used to "correct" for the
complication of more people see more birds and cover more area. If the
number of participants is not accurate, you are just increasing your error.

No data set is error free. That is why there is an entire science of
statistical analysis. When using any data set, it is important to think
about where the potential errors come from and the best way to correct for
those errors. If you don't believe me, try flipping a coin about 10 times.
You should come up with 5 heads and 5 tails but it can come up in many
different combinations.

As an aside, it is not uncommon for scientific publications to charge a per
page publication price. Many publications have really steep prices; some
times several hundred dollars per page. The $5 per head give the Audubon
CBC report the highest per page publication price Dr. Rintoul could find.
When he submitted his study for publication in American Birds, it was turned
down. I don't think he has tried to publish it again.

As I mentioned, the Kansas Ornithological Society publishes all of the CBC
for the state, free of charge, in their scientific publication. Many of the
longest running counts have never been submitted to the Audubon Society,
including the Manhattan count (which is one of the longest running counts in
the state). The average number of participants is 100.

Perhaps it is time to take the CBC to the computer age. Bypass the Audubon
Society and publish your CBC numbers on the web for anyone to down load. Or
we could publish all of them in Washington Birds for WA and Oregon Birds for
Oregon (and we could offer to publish Idaho and British Columbia CBC as
well). Of course, this would make it more difficult to look at continent
wide trends.

The current CBC database is a great resource. Just be careful how you use
it.

Deb Beutler
Dept. of Zoology
Washington State University
Pullman, Whitman Co., WA

dbeutler at wsunix.wsu.edu