Chimp & See Talk

Bias showing site name before classifying?

  • lauraklynn by lauraklynn moderator

    I am curious if there is any concern that people might bias their classifications based on having knowledge of the site prior to completing classification for each clip? I know a map does not show up until after classification is complete, but once you have completed a single video for a site, you have knowledge of where it is for all subsequent videos. In other words, each video shows the site name (e.g. lingering-shape) at the top and I know the approximate geographic location of the site from previous maps. Perhaps the location from the map is vague enough that this is unimportant? Are all the maps exactly the same?

    Posted

  • AnLand by AnLand moderator

    Hi @lauraklynn,

    the map you are seeing belongs to the clip beforehand. So, you actually see the map only after classification is finished. As at the moment only one site is open, there is of course knowledge you are using. But you are right, it could bias your answer and that's why you see the map only after you hit the finish button.

    Here is it explained briefly: http://talk.chimpandsee.org/#/boards/BCP0000002/discussions/DCP00009aw

    Posted

  • ksigler by ksigler moderator in response to lauraklynn's comment.

    Hi, @lauraklynn. What do you mean by bias in classification? As in, if I see something that could either be a gorilla or a chimp, but I know this site is in an area where there aren't any gorillas, I choose chimp? Yes, I suppose that is a type of expectation bias ("I expect only chimps here, so that's probably what it is"), but I don't think it's a bad thing when it's an application of one's knowledge of facts. The more we know, the better our classifications will be.

    Also, the classifications are very general, for the most part, not reaching down to the level where individual species' distribution will be too troublesome. For example, the 2 types of mangabeys we've seen so far. We've learned that agile mangabeys are in Central Africa, while sooty mangabeys are in Western Africa. In this case, knowing the site name/map of a clip could absolutely help us differentiate between the two. But when classifying, we only have "monkey" to choose from, anyway, so this extra knowledge doesn't come in to play (unless you want to add specific tags).

    There are probably lots of little biases, and other assumptions we make, during classifying. That's also why it's good that multiple people classify each clip, to help normalize it.

    Does this answer your question? I think it's very interesting to discuss the way we do this work and it's kind of a fun challenge to consider the variety of skills, experiences and motivations that volunteers have in a way that ensures this works for everyone. 😃

    Posted

  • AnLand by AnLand moderator

    Yes, I think too it can be helpful. For instance, at the current site Lingering Shape are many more black duikers than I've seen at other sites. (Which is cool, I love them!) We have also lots of unclear videos with big dark things behind branches, in twilight, and so on. The knowledge that this might be a black duiker (or: it is probably a black duiker) makes me look a bit closer before deciding whether I really classify chimp.

    The four regions A to D are in my opinion only the geographical range of the four chimp subspecies. So, at least for West Africa that is a huge area.

    Posted