Research Grade too easily obtained

In my view, it is too easy to get an observation to RG level without adequate reasoning. The following scenario happens exceedingly often:

User X uploads an observation which they cannot ID
User Y adds an identification
User X hits “Agree” but really has no personal knowledge to back that up.

A suggestion: Where a user initially uploads an observation and does NOT submit an ID, then RG should only be reached when two other users agree on an ID.

8 Likes

I also encountered such situations recently, mostly inexperienced users that might not yet have fully understood how iNaturalist is suppose to work.

For fairly common species this is ot that much of a problem. but for more rare species it might, if they go unchecked.

2 Likes

Yes, this is a real issue and there are “experienced” users who don’t get it (often, they see that a user they trust has made an ID and they confirm it; obviously they don’t know that person’s exact familiarity with the group). About 25% of the observations of the groups that I do are misIDed at first (usually by the computer guess). If I get to them first, I can get the right ID there (but realistically, that person shouldn’t just confirm my ID, either!). However, there are a lot of people that seem to just confirm whatever ID show up (often the original poster, sometimes others), and one (or sometimes two) of them will confirm the incorrect ID, sometimes in a few minutes. I assume some/most of these folks have personal knowledge to the point of genus, which is good, and but the attitude of “this must be the right species” which is bad. This of course, makes it RG. Since I know that there are a huge number of groups where people have not made a huge effort to clean the data, there are massive amounts of incorrect data going into GBIF etc. because of this problem.

4 Likes

Increasing the requirements for research grade makes sense to me, as an amateur who is definitely guilty of over-agreeing at times. Sometimes the scenario is that I start with a pretty good idea what I have but not quite enough to feel confident, so I go one taxonomic level above what I think I have observed, then agree if someone confirms my original thought. Other times I must admit it follows the exact scenarios others are describing here, even though I know it’s not supposed to work that way. I do try to be responsible but I think there’s a certain inevitability of succumbing to temptation/confirmation bias/etc.

There is also a middle ground where I know I lack the expertise, an expert IDs something for me, and then I do some web/field guide research on my own and agree if sources appear to support the ID. But, there are of course two problems with that as well: 1) I am not a trained expert and 2) I am starting the process with a presumed answer which may lend bias to my conclusion.

6 Likes

I know I have made these mistakes, would it be better for me to remove any IDs that I am uncertain of, that have not been ID’d by someone else?

What would be the best course of action to help improve RG?

For a past discussion on this subject, you may be interested in these threads:

8 Likes

No, I can’t see any need for that. It’s relatively unlikely for someone else to come along and agree to your ID, if yours was faulty.

1 Like

When I have agreed with someone else it is based on me making an effort to validate their ID. If users are required to document their findings in a truly scientific way, or by flashing their credentials, the casual user would be shut out. This is a “crowd sourced” app.
I look at “Research Grade” labels as having passed a basic straight face test and ready to validate for your research. If a tool designed to gather mass quantities of data was always right where would the fun be in data collection.

3 Likes

For some existing related feature requests, check out these:

When you have a specific feature request, please review the existing requests and submit a new topic in the #feature-requests category if yours hasn’t been requested yet. Thanks!

5 Likes

@acclivity. My experience has not bourne that out at all - faulty IDs at high levels are usually corrected pretty quickly, but more specific ones that are wrong (but close!) are confirmed constantly.

1 Like

I am a relative novice here who has tried to do more IDs as a way to increase exposure to species and to learn more - and yes, lockdown has been a factor in giving me some time. I commented on this issue in a discussion that flowed from the stats put up a couple of weeks ago regarding the site reaching a milestone number of observations.

I agree that new users might do as you describe - though we don’t know what sources were reviewed in the gap between the IDs. But sometimes, the scenario may be playing out as you suggest.

There may an upsurge in this because of the pandemic, new users, esp using the app?

To follow your suggestion above re users X and Y: broadly seems a good idea, but level of ID would need to be considered so with no ID at all, what you suggest sounds fair, and probably with allocation to eg flowering plants as well.

Another possibility is to have a pop-up at the point where that user (ie the one who posted the observation with either no ID, or with a high level ID only such as flowering plants) hits ‘Agree’, to say something like: “Please don’t just agree because someone else has proposed this ID. Can you positively identify this observation to this level for yourself? Yes? Then go ahead, agree!”

1 Like

I am such a novice at much of this that I don’t know how to reply to a specific user (meant to reply to @acclivity, nor indeed how to navigate all this and the significance of hashtags as in the post form the moderator @bouteloua Sorry!

1 Like

@wildlymistaken @hannahmc @steveplumb Welcome to the forum!

2 Likes

I am definitely enjoying collecting the data, as it were! I shall bear all of your comments & suggestions in mind when I find something that I’m not sure I can ID.

Thank you everyone

In an effort to consolidate a popular discussion topic and not duplicate conversations that have taken and may continue to take place I am closing this topic. If you have more to add, please consider adding posts in the existing topics linked by @whaichi and/or voting or commenting on feature requests linked by @bouteloua. Thanks.

This topic was automatically closed after 40 minutes. New replies are no longer allowed.