[obol] Re: eBird data entry - a warning (no sightings)
- From: Joel Geier <joel.geier@xxxxxxxx>
- To: Oregon Birders OnLine <obol@xxxxxxxxxxxxx>
- Date: Wed, 08 Oct 2014 10:40:01 -0700
Hi all,
The "Off-By-One-Line" ("OBOL" -- now there's a catchy acronym!) error is
likely much more common than most birders recognize. It's an inherent
problem with checklist-type data entry.
This type of error has long been the most common one to come up in
periodic reviews of www.birdnotes.net counts (what ebird folks call
"checklists"), so it's no surprise that it turns up in ebird too.
Let's not even get started on all of those "Omission Of Plain-jane
Species" ("OOPS"!) that surface when you circulate a draft compilation,
and someone notices that they forgot to record the day's only flock of
Bushtits.
The ebird robotic filters & volunteer reviewers can catch some of the
most egregious slip-ups, but it is wildly optimistic to expect them to
catch more than say, 1/4 of these types of errors.
As Mike Patterson described in a recent blog post, it is really
difficult to screen robotically for birds that are just a few miles out
of habitat, or where both species are present, but one is comparatively
rare (I think Mike's example was House Wren vs. Pacific Wren in certain
parts of Clatsop Co.).
The exercise that Stefan Schlick describes (producing a checklist of
birds for a favorite site, and then seeing if it makes sense) can
probably help to screen out a few more of these errors, but again, this
will probably only turn up the most glaring mistakes.
Organized counts that add more layers of human interaction & compilation
can help. In the CBC, where you sit around with your teammates and go
through your lists at the end of the day, usually there are a few
corrections during the countdown, and then a few more when the compiler
goes through the data, and passes the spreadsheet back to participants
for double-checking.
In compiling the checklists that birders recently turned in for the fall
migration count (NAMC), I've already come across several examples of
"OBOL" errors.
For example, a pair of "Red-shouldered" Hawks should have been
Red-tails. It was plausible habitat for either species. The only thing
that caused me to wonder was a footnote in a later count by the same
birder, suggesting that a couple of RTHAs might have been the same ones
as seen earlier on the route.
Heaven knows how many Black-throated Gray Warblers get recorded as
Townsend's Warblers or Hermit Warblers, and vice-versa, during spring
migration (never mind the difficulty of separating them by song).
A more robust form of data entry, rather than a checklist, would be an
app where you just type in the 4-letter banding code, and then confirm
that the bird name that pops up is the one that you expected.
I'm sure someone will tell me that such an app exists, but apparently
not everyone is using it. And that method will also have pitfalls. Not
everyone knows the 4-letter codes by heart (quick, what's the code for
Black-throated Gray Warbler?).
A confirmation step in such an app will help but let's face it, if
you're birding on the Coquille Valley CBC and you see a flock of gulls
heading up river while you're counting grebes and phoebes, you're
probably going to rush it a bit and will make some mistakes, either of
Off-By-One-Letter variety (OBOL again), or OOPS.
So yeah, no matter what your doctor tells you, an extra sprinkling of
salt is always good when you're trying to swallow "citizen-science."
Think of it as Irish stew -- it'll fill you up if you're hungry, but
just don't expect haute cuisine.
Happy counting,
Joel
--
Joel Geier
Camp Adair area north of Corvallis
OBOL archives: www.freelists.org/archive/obol
Manage your account or unsubscribe: //www.freelists.org/list/obol
Contact moderators: obol-moderators@xxxxxxxxxxxxx
Other related posts:
