Yes, that was my thought. It would be difficult to know the sex (or the gender) of an author name on a paper. There would inevitably be a lot that you could not determine. And certainly in the sciences multi-author pages are the norm and even where you did know the sex/gender of all, do you assign some part-score? E.g. 0 for all male, 1 for all female, 0.6 for 3 women and 2 men. But I am curious why you are asking the question? That the writing/research of women is under-represented in Wikipedia citations? If so, without conducting any research, I'd say "yes it is under-represented". But my reason would be because women are under-represented as writers/researchers in the first place. And certainly the older the source, the more likely it is to be written by a man. So to investigate gender bias in citations in Wikipedia, you would have to estimate the proportion of men/women (or at least their outputs) over time in a given discipline and then ask the question, "taking into account of the time of publication of a citation and the proportion of men/women active in this discipline at that time, do Wikipedia citations show a sex/gender basis?". Hmm ... very tricky. I'd be inclined to suggest starting with a much simpler task. Pick a discipline (preferably one with a professional society who can tell your their estimate of current male/female ratio over (say) the past 5 years), limit the Wikipedia articles to topics in that discipline, and limit the citations to those published within the last 5 years. Indeed, perhaps limiting it to publications that are principally from the same country(s) as the professional society from which you get the data (as clearly men/women's participation in any discipline can vary with different countries for cultural reasons). Then you have some way to gauge whether Wikipedia is showing more or less gender bias in its citations than the discipline itself exhibits through publication. Quite a challenge! And of course, it is not Wikipedia that adds citations. It is individual contributor who add citations. Does the sex/gender of the contributor have any correlation to any observed bias? Again, the task is made more difficult because a lot of Wikipedians don't identify their sex/gender. The other thing to be alert to is the difference in how (I believe) Wikipedians cite compared to researchers. As a researcher, I will of course be reading papers in my field all the time and what I read will influence my subsequent work. Therefore when I write about my research, my citations are referring to papers that I have already read and whose authors may be familiar to me from their other work, having met them at a conferences, private correspondence, etc. However as a Wikipedian, I am only partially operating that way (mostly when I write new articles or significantly expand them, that is, when I am doing the research). A lot of the time I am adding citations relating to content other people (often new users) have added/changed without citation. These come up on my watchlist all the time. What do I do? Of course I could revert saying "no citation provided", but that's not the way to encourage new contributors nor to grow the encyclopedia, so if the information seems plausible (not obviously vandalism), I will attempt to find a citation for it (using tools like Google and other topic-specialise search tools). This is what I call "lucky dip" mode of citing as obviously I have no idea what the source was for the original contributor. The sources I find from my search may not already be known to me (frequently they are not). Or to summarise, IMHO, researchers (or Wikipedians in "new content mode") cite a source already known to them and whose authors may be known to them and could consciously or unconsciously engage in some discrimination in citation based on sex/gender or other criteria, whereas Wikipedians in "updating mode" are likely to be citing a source not previously known to them and may be happy just to have found a source and are unlikely to be spending a lot of their time researching the authors of that source to be extent they could then consciously or unconsciously exercise discrimination on sex/gender. If I invest any extra effort in such a situations, it's probably because the wording of the source is a close match to the Wikipedia article which begs the question of copyright violation (which needs to be dealt with by deletion or rewriting) or being a Wikipedia mirror (which is obviously not an acceptable citation). So I suspect whether a citation was added by the same contributor as the content it supports or a subsequent contributor probably makes a difference to the likelihood of conscious/unconscious discrimination. Also, finally, often Wikipedia cites web pages and other sources that do not have any individual authorship, e.g. government websites. Remember that Wikipedia prefers open citations over paywalled citations and a lot of the publications behind paywalls are individually authored. Your proposed research has a lot of interesting challenges and a number of limitations. I'm not saying don't do it, but I am saying start very small and see if you can find any evidence to support your hypothesis before embarking on a larger study. Because contributor behaviour is what you are trying to study, you probably need to do both quantitative and qualitative experiments. E.g. I have described the two modes of citation I do, but I cannot say how typical my behaviour is. Kerry -----Original Message----- From: Wiki-research-l [mailto:wiki-research-l-bounces@lists.wikimedia.org] On Behalf Of Leila Zia Sent: Friday, 23 August 2019 3:44 AM To: Research into Wikimedia content and communities <wiki-research-l(a)lists.wikimedia.org> Subject: Re: [Wiki-research-l] gender balance of wikipedia citations Hi Greg, A few comments if you're going to go with "proportion of male vs female authors of the source material used as citations in arbitrary articles": * Please differentiate between sex (female, male, ...) and gender (woman, man, ...). My understanding from your initial email is that you want to stay focused on gender, not sex. * Unless you have reliable sources about the gender of an author, I would not recommend trying to predict what the gender is. (As you may know, this is not uncommon in social media studies, for example, to predict the gender of the author based on their image or their name. These approaches introduce biases and social challenges.) * Re your question about whether WMF has resources to look into this question in-house: I can't speak for the whole of WMF, however, I can share more about the Research team's direction. As part of our future work, we would like to "help contributors monitor violations of core content policies and assess information reliability and bias both granularly and at scale". [1] The question you proposed can fall under assessing bias in content (considering citations as part of the content). I expect us to focus first on the piece about violations of core content policies and information reliability and come back to the bias question later. As a result, we won't have bandwidth to do your proposal in-house at the moment. Sorry about that. I hope this helps. Best, Leila [1] Section 2 of our Knowledge Integrity whitepaper: https://upload.wikimedia.org/wikipedia/commons/9/9a/Knowledge_Integrity_-_W… On Thu, Aug 22, 2019 at 9:57 AM Greg <thenatureprogram(a)gmail.com> wrote: _______________________________________________ Wiki-research-l mailing list Wiki-research-l(a)lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wiki-research-l