Hi Piotr, I think this may be due to gaps in the underlying data.
From memory, Wikidata initially showed higher proportions of women than was expected from what we knew about Wikipedia, because the existing category structure (eg "Female X" but not "Male X") meant that more of the "obvious" cases, which could be inferred from categories, were women. This then dropped down a bit as more of the "less obvious" data was added. *If* this is generally true, and I believe it is, then projects with a high proportion of biographies which do not have gender set will be more likely to have inaccurate gender statistics - because the obvious cases get added first, but these are not generally representative. Very cursory check, looking for all P31:Q5 with sitelinks to given projects: * zhwiki has 132k known people, of which 56k do not have P21 set. * jawiki has 252k known people, of which 79k do not have P21 set * frwiki has 424k known people, of which just 94(!) do not have P21 set * eswiki has 248k known people, of which 76 do not have P21 set So 42% of biographies on the Chinese Wikipedia, and 31% on Japanese, do not have known wikidata gender. Meanwhile, the proportions without gender on Spanish and French are so low as to be negligible. These numbers also indicate a very large proportion of biographies on these projects don't overlap with Western language coverage (or Wikidata would have picked up gender from those). Given what we already know about how articles are distributed across projects, it seems reasonable to assume they mostly match to people from within the relevant language's cultural group - and there, perhaps, is your explanation... Regards, Andrew. On 16 January 2015 at 11:49, Piotr Konieczny <[email protected]> wrote: > We are very happy to hear that your findings corroborate ours. What's your > take on what we think is rather surprising high percentage of female bios > for the South Asian/Confucian clusters? (i.e. non-Islamic Asia)? > > -- > > Piotr Konieczny, PhD > http://hanyang.academia.edu/PiotrKonieczny > http://scholar.google.com/citations?user=gdV8_AEAAAAJ > http://en.wikipedia.org/wiki/User:Piotrus > > On 1/13/2015 22:07, Magnus Manske wrote: > > To spam this list as well as Twitter :-) > > http://magnusmanske.de/wordpress/?p=250 > > On Tue, Jan 13, 2015 at 5:41 PM, Maximilian Klein <[email protected]> wrote: >> >> Thank you all for the feedback. I will have taken away quite a few good >> ideas for further investigation, to summarize: >> >> Gerard - look at the ratios of those bios of a language, which exist only >> in that language. >> Han Teng - "male gaze" hypothesis, create a by-profession crosstabular >> analysis. >> Jane - look at the ratios of leading actors by language, and "fictional >> humans" more closely. >> Jonathan - perform a filter step, or perhaps a weighting by page-views. >> >> Thanks so much for the advice, what a great list. >> >> Make a great day, >> Max Klein ‽ http://notconfusing.com/ >> >> On Mon, Jan 12, 2015 at 5:57 AM, WereSpielChequers >> <[email protected]> wrote: >>> >>> I have spent quite a bit of time at new page patrol over the years. My >>> suspicion is that many if not most of the people who create articles on >>> newly signed pop stars and actors are from their management agency rather >>> than fans, especially if they seem too early in their career to have fans. >>> Sportspeople I suggest are more likely to be written about by fans, >>> especially if they have been signed by a major team, or more importantly for >>> Wikipedia a team with an actively editing fan. >>> >>> On this theory the quality of articles, the number of edits, and when we >>> had the Article Feedback Tool the number of "is hot" type comments would be >>> a good indication of interest from the volunteer editing community. But >>> article creation is in part a matter of the policy of the relevant talent >>> agencies. >>> >>> Sorry if that sounds overly cynical, perhaps if it were possible one >>> would filter out the articles that get scarcely any views and then look at >>> the gender balance of articles that are of interest to our audience as well >>> as our editors. >>> >>> Regards >>> >>> Jonathan Cardy >>> >>> >>> On 11 Jan 2015, at 22:23, h <[email protected]> wrote: >>> >>> Hello Piotr and Gerard, >>> >>> I think a competing hypothesis would be "male gaze". That is to say, >>> the more female representation is not about a culture (defined as national, >>> ethnic, linguistic or regional, not macho/feminine), but rather a >>> gender-interest bias. Thus the more female representation could mean more >>> male dominant culture, which is against the theoretical assumption of >>> Piotr's research. >>> >>> Note that East Asian Wikipedians that I know, especially those who >>> edit Chinese Wikipedia, are predominantly very young. Some of them can be >>> highly interested in opposite sex. >>> >>> Check the following category pages as examples: >>> (1a) Female actresses of every countries in the world >>> >>> http://zh.wikipedia.org/wiki/Category:%E5%90%84%E5%9C%8B%E5%A5%B3%E6%BC%94%E5%93%A1 >>> (1b) Male actresses of every countries in the world >>> >>> http://zh.wikipedia.org/wiki/Category:%E5%90%84%E5%9B%BD%E7%94%B7%E6%BC%94%E5%91%98 >>> >>> (2a) Female Japanese AV (i.e. porn) actresses >>> >>> http://zh.wikipedia.org/w/index.php?title=Category:%E6%97%A5%E6%9C%ACAV%E5%A5%B3%E5%84%AA >>> (2b) Male Japanese AV (i.e. porn) actresses >>> >>> http://zh.wikipedia.org/w/index.php?title=Category:%E6%97%A5%E6%9C%ACAV%E7%94%B7%E5%84%AA >>> >>> It is quiet clear that the male gaze hypothesis seems to apply here. >>> More female presentation simply because they are there to be consumed by men >>> or boys. >>> >>> So one of my suggestions for research is to select a few professional >>> categories that are of interest (say, politicians, poets, entertainers, >>> etc.) to do some cross-tab analysis. >>> >>> Thus, I will be extremely cautious against using the current >>> metrics/methods as viable "gender inequality index". >>> >>> As a proponent of "data normalization" and "geographic normalization" >>> method myself, I would distinguish two sets of comparisons: one is >>> cross-country or cross-language version absolute value comparison, another >>> is cross-country or cross-language version "normalized" value comparison. By >>> geographic normalization, I mean that researchers must gather another set of >>> cross-country or cross-language datasets that captures some aspects of >>> realities "external" to Wikipedia. In this case, I would say the Wikipedia >>> represented politicians' gender ratio against the offline gender ratio of >>> politicians. In other words, "data normalization" allows researchers to >>> compare which language version are more or less (and how much) equal than >>> the corresponding offline societies. >>> >>> BTW, the methods you develop to extract gender from biography >>> articles for large-scale analysis may also be re-purpose to study other >>> dimensions. One dimension that will interest me would be nationality. It >>> will be interesting to see the coverage, focus or bias of a language version >>> on people based on nationalities. Age might be another one. >>> >>> Best, >>> han-teng liao >>> >>> >>> >>> 2015-01-11 19:01 GMT+02:00 Gerard Meijssen <[email protected]>: >>>> >>>> Hoi, >>>> Having read it, I find it is still very much a Wikipedia oriented.It >>>> makes use of the toolset by Markus. That is fine. the notion of diversity >>>> and notability is also very much culturally defined. It would be nice to >>>> know how the different wikipedias accept notability of people from other >>>> cultures and if it impacts the diversity of their own articles. >>>> >>>> I have found that many people do not have an article in the languages of >>>> their own cultures. Often it has to do with an interest in a domain that is >>>> more of relevance to the other culture. >>>> >>>> Diversity is very much part of a domain; in Roman Catholicism male >>>> dominance is obvious. I am curious if diversity in gender is affected by >>>> such considerations and if items with a single article are more in line >>>> with >>>> what is the norm for a culture, a domain. >>>> Thanks, >>>> GerardM >>>> >>>> On 10 January 2015 at 11:51, Piotr Konieczny <[email protected]> wrote: >>>>> >>>>> Here >>>>> (http://notconfusing.com/preliminary-results-from-wigi-the-wikipedia-gender-inequality-index/) >>>>> are some early findings from a research project I am involved in (together >>>>> with Maximilian Klein). (To find out more about the project, see >>>>> https://meta.wikimedia.org/wiki/Research:Wikipedia_Gender_Inequality_Index >>>>> and it's talk page). We are very curious what you think (don't hesitate to >>>>> be critical). What we would really appreciate would be any alternative >>>>> hypotheses (to the one presented) that could try to explain why post-1950s >>>>> Confucian and South Asian clusters seem so much more inclusive of female >>>>> biographies than others (including the "Western" clusters). Are we seeing >>>>> a >>>>> data error, or something else - and if so, what? >>>>> >>>>> -- >>>>> Piotr Konieczny, PhD >>>>> http://hanyang.academia.edu/PiotrKonieczny >>>>> http://scholar.google.com/citations?user=gdV8_AEAAAAJ >>>>> http://en.wikipedia.org/wiki/User:Piotrus >>>>> >>>>> >>>>> _______________________________________________ >>>>> Wiki-research-l mailing list >>>>> [email protected] >>>>> https://lists.wikimedia.org/mailman/listinfo/wiki-research-l >>>> >>>> >>>> >>>> _______________________________________________ >>>> Wiki-research-l mailing list >>>> [email protected] >>>> https://lists.wikimedia.org/mailman/listinfo/wiki-research-l >>>> >>> >>> _______________________________________________ >>> Wiki-research-l mailing list >>> [email protected] >>> https://lists.wikimedia.org/mailman/listinfo/wiki-research-l >>> >>> >>> _______________________________________________ >>> Wiki-research-l mailing list >>> [email protected] >>> https://lists.wikimedia.org/mailman/listinfo/wiki-research-l >>> >> >> >> _______________________________________________ >> Wiki-research-l mailing list >> [email protected] >> https://lists.wikimedia.org/mailman/listinfo/wiki-research-l >> > > > > _______________________________________________ > Wiki-research-l mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/wiki-research-l > > > > _______________________________________________ > Wiki-research-l mailing list > [email protected] > https://lists.wikimedia.org/mailman/listinfo/wiki-research-l > -- - Andrew Gray [email protected] _______________________________________________ Wiki-research-l mailing list [email protected] https://lists.wikimedia.org/mailman/listinfo/wiki-research-l
