Thursday, November 28, 2013

The Accuracy of eBird and the Precision of eBirders

I have a love/hate relationship with eBird. I love it because it is awesome, but it is full of people and I hate people. No, that isn't quite fair since there are plenty of people I don't hate, but any endeavor that involves people, especially interacting people, is bound to have annoyances.

This is a common theme here on The Year of Birds, so much so that we have a tag for "other birders", but it might as well be called "other people". Also common is when we whine a little about eBird, but that is more so because it does so much well that it is glaring when things aren't so good. One thing it is good at, though, is giving you lots information about a birder just from look at the lists they take.

For example, let's say you see a birder with a pretty rare sighting and want to know how good a birder they are? Click on the list that that the rare bird is on, it'll give you lots of little subtle information. What things set off alarm bells for me?

-"Gull sp". Now, I know sometimes there can be hundreds of gulls in the distance, but I can't tell you how many times I've seen "gull sp." listed on a list from an area where gulls are pretty close. A listing of "gull sp." might indicate that they were far away, or it might indicate that the birder doesn't know basic gull differences, or it might indicate indifference. This isn't damning evidence, but if I'm skeptical already about an identification, this doesn't help.

-A low number of species in a birding hotspot despite being there for many hours. I was at a new hotspot for me recently and got lots of birds there easily. Later I saw another person spotted a rare bird there and was there for a longer period of time, but had literally 1/3 the birds I had. I mean, you have to work hard to not get that many, or you aren't that good of a birder and your ID of a rare bird may be suspect. Luckily, the rare bird was a continuing and obvious bird, but if they were the only one seeing it, I'd wonder.

-Lots of "X" numbers on species instead of actual numbers. This annoys me because it gives almost no information about the birds they saw. It seems to me that if they aren't to be bothered to give even a rough estimate on the number of crows they see, why would I think they put the effort into identifying a rare bird?

-Not having any detail. Now, don't get me wrong, I'm guilty of not providing field marks on rare birds all the time, but I put down some notes at least. If you aren't even saying where it was or anything else, it is hard to believe you. Of course there are plenty of long-time birders around whose ID I'll take on their word.

-Improbably precise number on large groups of moving birds. Ughh, this is the worst. I'll give an example without much detail (to not call anyone out). Earlier this year I saw someone report a Black-capped Chickadee at Heinz, and the description left me very skeptical, so I looked at the whole list. In the list it had unlikely precision on ducks, such that there were something like 83 Mallards, 136 Green-winged Teal, 119 Northern Shovelers, etc. (the number are not the same to protect the innocent). That set off red flags for me.

If you say you saw 136 Green-winged Teal, that has a strong implication of being an exact count. If you said 130, that would imply (because we use base 10) a range around 130 (say, between 125-135), which despite being less precise, would actually be more accurate because birds move around (even a little bit) and people make mistakes, especially when walking around for a lake for a few hours. But any number that is like 222 or 931 becomes something that I immediately think of as amateurishly inaccurate.

I once saw a list of birds that listed on bird as having 236 because they saw 36 in one area and estimated 200 at another area. Ughh, you may be perfectly precise with your 36, but once you have an estimate in there, your WHOLE COUNT is an estimate, and your numbers should reflect it. 240 would have been a better number there.

Apparently the Black-capped Chickadee sighting wasn't confirmed, and that was probably due to email correspondence asking for better description. But I wonder how much the eBird people look at the other things on the list to give them an impression of how accurate an ID is?

No comments:

Post a Comment