Is Most Likely good enough?

I marked a species observation to the genus because the photo presented does not provide enough data (and really can’t unless in the hand) or by voice. Voice was not presented but the person said they heard it and it is most likely true. It may be accurate but the evidence here is just not there. Friends, apparently, then supported the first report - again, without further evidence.
At best it should be marked by genus but short of that, as Casual.

Built in errors like this must skew the data at some point.


The additional evidence is there if the person provided it, even if only in text-form. The convention is to assume people aren’t lying (until shown otherwise). I’m not sure what you mean by “voice”, but I assume you mean a sound made by the organism? Did the observer describe the sound or did they just say something like, “it sounded like X”?


It is a bird that can only be identified by call/voice or in hand. They said they recognized the call. Most experienced birders can do that but iNat has only photos or recordings or it is considered Casual.

To me, it comes down to whether they provided evidence of the sound they heard (e.g. describing the distinctive sonic features of the bird-call) instead of a conclusion (e.g. “it sounded like bird species X”). People supplement their photo- or sound-observations with important field notes all the time. You should not be marking all such observations “casual”.


All citizen science data, even all scientific data, contains errors. Good researchers understand this, so it’s no big deal.

[Edit added later: but, of course, we should avoid errors and strive to fix them whenever we can–it’s just not worth obsessing over is all I’m saying]


I think a lot of this comes down to identification style. At one end, there’s being very reserved and only suggesting an ID if there are enough features to rule out anything else, including that one rare species last seen in 1910. At the other, there’s willingness to ID assuming that nearly all the others in the area were species X, therefore this one is also species X. I think identifiers fall somewhere on this spectrum, often in different places, depending on genre, family, etc.

There’s also beginning identifier confidence (see: Tomatoes, berries, fruits, and vegetables - discuss!), where awareness of the rarer species may be lacking, leading to overconfident IDs.

In such cases where you feel like the ID is overconfident, you can ID it to a higher level (less specific) without disagreeing, but leave a comment stating what would be needed to get it to the lower level (more specific).

Alternatively, if it’s something like a Gerald (see: Happy Gerald Day), you can tag in (@) some of the other identifiers you interact with to help disagree with an unevidenced identification.


In these cases, if the observer says that they did actually differentiate it then I tend to trust them, and I’ll put a non-disagreeing higher taxon ID. E.g. if they say they identified something by the call, or they post a blurry photo and say something like “sorry for the poor photo but it was definitely X”. In these cases I trust that the observer is correct, but I personally can’t support their ID from the evidence provided.

If there’s no indication that their ID is anything more than a guess, then my go-to is to put a disagreeing higher taxon ID with a comment like “how did you differentiate it from XYZ?”. That way it gives them a chance to explain their reasoning and I can evaluate it when/if they reply.


In my view, an observation should be independently verifiable – at least in principle. Otherwise it does not deserve an RG status (unless it’s a lowest verifiable taxon).


Right, but isn’t it reasonable to reduce the number of possible errors? :)


Absolutely! But there are hundreds of thousands of errors…and we can’t fix them all. We ID as best we can and then move on. I’m just saying that we don’t need to lose sleep over each one. ID to a reasonable level of confidence using the evidence that can be observed.


I think the issue is in the others that confirmed the ID for Research Grade. Nothing wrong with someone stating an ID from their own evidence, but other people should not be confirming it if they have none of that additional information available to them.

But we’re not snobs, so people make mistakes and we all do it our own way within the rules provided.


I think I would want all required ID information present in the post for something rare or unlikely, but if it was, say, a Western Meadowlark in WM habitat, good enough seems fine.

It depends also on your experience of that observer. Say two birders who regularly walk together, and know and trust each other’s bird ID skills. I will pull birds from Unknown, and leave the birders to sort it out.

You can agree at the higher level where you are confident. Or mark as reviewed and move on?

I would think an experienced birder submitting their field notes saying what the bird sounded like should be considered good evidence for an ID. Especially if it’s a relatively easy call for an experienced birder to make, like American vs. Fish Crow or Eastern vs. Western Meadowlark. I expect the rate at which a lifelong birder would misidentify a clearly-heard textbook crow call in the field is on par with the rate at which people accidentally upload photos with the wrong date or location. We all trust people’s location and date data without asking for verifiable evidence that the photo was taken at the claimed location, and I see trusting their statement that “this crow was making obvious fish crow noises 10 feet in front of me” the same way. Would marking every observation without EXIF data to prove where it was taken as Casual eliminate errors? Sure, but it would throw the baby out with the bathwater, so to speak. IMHO, ditto with questioning everyone’s field notes. I think the cost of letting in an occasional error by trusting field notes is worth the plethora of “almost certainly correct” RG observations that we get by doing so.


I’m not sure why it should be marked as ‘casual’ f it can be confirmed to genus. Instead, consider identifying to genus level - with a disagreeing identification if necessary - and using the Data Quality Assessment (DQA) to indicate that the Community Taxon cannot be further improved. That process can result in the observation achieving research grade at the genus level – placing a green ‘research grade’ label on the observation while avoiding over-specification.


All science is based on the principle of “most likely.”

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.