[obol] Re: Credibility (shorter & less rambling version)
- From: Joel Geier <joel.geier@xxxxxxxx>
- To: Oregon Birders OnLine <obol@xxxxxxxxxxxxx>
- Date: Mon, 07 Jul 2014 10:35:36 -0700
Hi again all,
Looking at what I wrote last night, it seems to be a good illustration
of why it seldom pays to start writing something after normal bedtime.
Here is an attempt at a more concise version, which will hopefully make
better sense:
My suggestion is that the current shape of the eBird review process,
with its strong focus on "rarities" (whether in time or space), is
mainly an outcome of an uneasy marriage between science (the effort to
gain knowledge about bird distributions etc.) and competitive birding (a
sport in which participants compare their accomplishments in terms of
"big day" lists, "big year" lists, etc.). Again, I am sympathetic
because we had to deal with the same issues concerning BirdNotes.
Let's start with the premise that eBird, like earlier similar projects,
was set up with the aim to collect casual bird observations by birders,
with the hope of gaining useful information about bird occurrence
patterns.
Many of the most active birders in a given region tend to be
list-oriented. The most common types of lists, apart from yard lists,
tend to be based on politically defined regions such as states,
provinces, or counties. Hence the most active birders tend to be keenly
aware of the status of birds in the states or counties where they
regularly go birding.
It is arguable whether rare-bird reports have any significant impact on
the scientific aims of data-gathering projects such as eBird. For a
robust scientific analysis of data, one normally excludes "outliers."
Significant findings (in terms of timing of migration, breeding range
extensions, etc.) generally need to be supported by a large number of
observations, not just one or two unusual reports.
For example, it would be ill-advised to re-draw the range map for Common
Yellowthroat just because someone reported one in Wheeler County,
regardless of whether that observer is considered "credible" or not by
their peers in the birding community. However, if multiple observers
start to see a species with regularity during nesting season (as we've
seen in the case of Red-shouldered Hawks in recent years here in Benton
County), then there is robust support for changing the range map. If a
trend is real, eventually good observations will swamp the anomalies.
That approach works OK in a purely scientific effort. The problem is
that eBird (like BirdNotes before it) is trying to glean scientific data
from a recreational activity, in which the leading participants --
list-oriented birders - tend to pay lots of attention to outliers,
a.k.a. "rarities."
When "rarities" (I'm using quotes here because most "rarities" are
common someplace else) show up in a database, list-oriented birders tend
to take notice. If they feel that some of these "rarities" were
incorrectly identified, they start to criticize the data gathering
effort. We saw this with BirdNotes, and we're still seeing it with
eBird.
If a database loses credibility among these birders -- who tend to be
among the most prominent birders in a given state -- their views might
discourage other birders from participating in the data gathering
effort.
Hence even if "rarity" reports are usually not significant for the most
credible types of scientific analysis that could be done using data from
recreational birding, they can have a big impact on how the database is
viewed by birders. This provides motivation to give special attention to
"rarity" reports, far out of proportion to their actual scientific or
conservation significance.
For the sake of science and bird conservation, I suggest that a
different focus to the review process would be desirable. Even if it's
still necessary to flag county-level rarities in order to keep up
appearances with list-oriented birders who are keenly alert to any
surprises in their favorite patch of real estate, why not put at least
equal focus on birds that are of conservation concern?
For example, when I've looked up Vesper Sparrow records in western
Oregon (where the nesting subspecies, Oregon Vesper Sparrow, has been on
the state list of Species of Concern for many years), I seldom see any
details that I would expect if these records were being reviewed. For a
population that seems to be well below 2000 birds (and falling),
incorrect identifications could have a significant impact on the
picture that emerges from eBird. Beginning and even intermediate birders
could easily mistake a Song Sparrow for a Vesper Sparrow, and even
advanced birders could sometimes get fooled when trying to identify one
by ear (for example, if you hear a distant Bewick's Wren or Song
Sparrow).
Again, I think that censoring data would be a mistake -- this introduces
its own type of bias in the dataset. Rather, I would like to see birders
being encouraged to clarify the basis for their identification of
species of conservation concern: Was the bird seen, seen and heard, or
heard only? And how well was it seen or heard? This type of information
can be useful for following up reports, to try to confirm whether a
species of concern is using a given patch of habitat.
I think if birders see more focus on conservation issues, and less on
"rarities," there could also be a positive effect on data gathering:
more focus on correctly identifying birds for which a few data points
could really make a difference, and less focus on trying to list more
species for the year/month/county/state whatever.
Good birding,
Joel
--
Joel Geier
Camp Adair area north of Corvallis
Other related posts:
