[NatureNS] Call count survey results

DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;
From: Bob Farmer <farmerb@gmail.com>
Date: Thu, 22 Mar 2012 12:00:17 -0300
To: naturens <naturens@chebucto.ns.ca>
Precedence: bulk
Return-Path: <naturens-mml-owner@chebucto.ns.ca>
Original-Recipient: rfc822;"| (cd /csuite/info/Environment/FNSN/MList; /csuite/lib/arch2html)"

next message in archive
no next message in thread
previous message in archive
Index of Subjects


Dear all:
A few years ago, I invited readers to participate in a web-based
survey of birdsong identification abilities that I was conducting with
Marty Leonard and Andy Horn.  I received a great response and lots of
support, and the results of that research have since been published in
the journal The Auk (University of California Press).

Although we didn't make the project goals explicit ahead of time (this
was to keep from biasing the results), we wanted to see what tends to
happen when birdwatchers listen to birds that sound quite similar, for
instance the Rose-breasted Grosbeak and American Robin, but where one
of these birds might be more rare (more "sexy") than others.  We were
curious how an outside pressure to hear rare species might bias these
kinds of detection "decisions".  We also wanted to see how reliable
observers can be when they say that they're "confident" they got
everything right.

What we found was that as self-reported skill levels increased, missed
detections and false-positive detections were less frequent (no
surprises there).  What was interesting, however, was that when expert
birdwatchers made false-positive errors, they tended to report a rare
species.  On the other hand, less-skilled birdwatchers ("moderates")
tended to report more common species when they made such mistakes.  A
mild incentive to detect rare species (a higher "score" in this case)
didn't seem to affect these patterns.

We also found that when the bar is set really high for confidence (as
in, "did you get everything right?  yes or no?" for a very difficult
survey scenario), observers of all skill levels tended to be poor
judges of their own performance (in other words, the amount of
overconfidence is pretty consistent).

Since some amount of detection error is basically unavoidable in bird
call surveys, these results will make survey designers think more
carefully about the skill levels of their observers, because the
nature of these errors can vary with the observer's skill (e.g. more
false positives of rares or commons), and it's not good enough to
trust observers to self-rate their imagined accuracy.  On the other
hand, these early results don't indicate that some healthy competition
to find the rare birds is going to bias things.

Thank you very much for your support, and I hope your group finds
these results interesting.

Sincerely,
--Bob Farmer
PhD candidate, Dalhousie University (Halifax, NS, Canada)
http://leonardlab.biology.dal.ca/Bob.html

ps:  The article can be accessed at http://dx.doi.org/10.1525/auk.2012.11129
If you're unable to access the full text would like to see it, please
write me and I can provide a copy!
Its formal citation is:
Farmer, R. G.; Leonard, M. L. & Horn, A. G. (2012). Observer effects
and avian call count survey quality: rare-species biases and
overconfidence. Auk. doi:10.1525/auk.2012.11129

next message in archive
no next message in thread
previous message in archive
Index of Subjects