Wikipedia:WikiProject Video games/Newsletter/20160106/Feature

Feature: Gender Declaration and the WP:VG Gender Gap
 Submitted by Thibbs 

To anyone who spends any time reading about video games as a socio-cultural phenomenon, it will come as no surprise to hear that for decades the population of male gamers has outnumbered the population of female gamers. To anyone who spends any time reading about Wikipedia as an institution and about its continued relevance nearly a decade and a half after its founding, it will come as no surprise to hear that the population of male Wikipedians has been found to outnumber the population of female Wikipedians. Multiply the fractional number of female gamers by the fractional number of Wikipedians and we find ourselves with WikiProject Video Games: an 11-year-old user group that ranks among Wikipedia's largest and most active editing collectives, and whose population of openly female editors numbers in only the single digits.

The Gender Gap issue and its related partner, Systemic Bias, have steadily risen in prominence since the end of the 2000s, and related issues of harassment and civility have become a mainstay of discussion and argument at Wikipedia's many "drama boards" (up to and including ArbCom with recent rulings in WP:ARBGGTF, WP:ARBGG, WP:ARBAE2, and the earlier PFC Manning naming dispute). Indeed, at the recent ArbCom elections that took place last month, nearly half of the WP:ACE2015 roster referred to the gender gap or systemic male-oriented bias in their candidate statements, with another half discussing the "GamerGate" situation (covered in the newsletter last quarter) by name in their candidate questions. Apart from GamerGate itself (and its related articles), there has not been much in the way of gender-related disagreements within WP:VG, however the lack of drama may only underscore the lack of gender diversity within WP:VG's editing population.

The WP:VG gender gap may not come as a surprise to readers who are aware of their surroundings, but whether the WikiProject in fact represents an outlier when compared against other WikiProjects may not be as obvious. Also not immediately obvious are the reasons underlying the gap in both gaming and Wikipedian populations, and what (if anything) can be done about it. This quarter the Newsletter takes a brief look at the scope of WP:VG's gender gap, examines how it has changed over the years, and provides a cross-WikiProject comparison to contextualize the WP:VG gap with the gaps at some of Wikipedia's other largest WikiProjects.

Methodology
The cross-WikiProject study was intended to encompass the largest WikiProjects, and as such it has somewhat of a bias toward older and more stable projects. Largest WikiProjects were determined by drawing "Top 10" lists from six different lists of WikiProjects providing five size metrics. Specific factors taken into account include:
 * 1) Number of articles stewarded (2011) and (2015) - Both datasets were included in order to emphasize long-term size
 * 2) Number of main talk page watchers
 * 3) Number of participants (members, contributors, etc.)
 * 4) Number of edits - Number of edits by any user to stewarded articles
 * 5) Number of editors - Number of unique editors to stewarded articles (i.e. number of edits less those made by the same editor)

WikiProjects appearing on multiple lists were slated for deeper analysis, with priority generally based on the number of lists hosting the WikiProject, and secondary priority based on the topic's relationship to gender. The final list of WikiProjects subjected to analysis includes:
 * 1) WikiProject Military history (WP:MILHIST)
 * 2) WikiProject Football (WP:FOOTY)
 * 3) WikiProject Film (WP:FILM)
 * 4) WikiProject Biography (WP:BIO)
 * 5) WikiProject United States (WP:USA)
 * 6) WikiProject Medicine (WP:MED)
 * 7) WikiProject India (WP:INDIA)
 * 8) WikiProject Video games (WP:VG)
 * 9) WikiProject Deletion sorting (WP:DS)
 * 10) WikiProject Albums (WP:ALBUM)
 * 11) WikiProject Television (WP:TV)
 * 12) WikiProject Articles for creation (WP:WPAFC)
 * 13) WikiProject Mathematics (WP:WPMATH)
 * 14) WikiProject Spam (WP:WPSPAM)
 * 15) WikiProject LGBT studies (WP:LGBT)
 * 16) WikiProject Women (WP:WOMEN)
 * 17) WikiProject Countering systemic bias (WP:CSB)
 * 18) WikiProject Feminism (WP:FEM)
 * 19) WikiProject Gender Studies (WP:Gender Studies)

Membership rolls from the last day of each year following the founding of each WikiProject were queried using the "gender" magic word or the Wikipedia pronoun template, and numbers of editors in each of the three categories (m/f/u or he/she/they) were recorded. Percentages of female membership were determined by dividing the number of declared females by the total number of declared editors (i.e. the total excluding undeclared editors) and results were graphed chronologically on the same population growth axes as the other WikiProjects under consideration.

Declared editor totals were later compared against undeclared editor totals and representatives of each of the resulting gender growth categories were graphed to reveal possible connections between the number of undeclared editors and the gender ratios in each gender growth curve.

Chronological comparison of WikiProject gender statistics
The above images marked "Low scale" (at left) and "High scale" (right) present information concerning the number of female WikiProject members as a percentage of the total number of gender-declaring editors, and not as a percentage of the total number of members. This is a significant factor for these graphs as the number shown is unlikely to reflect the true percentage of women participating in each of the listed WikiProjects. With gender-undeclared editors making up anywhere from two to five sixths of the members of any of the WikiProjects considered in this study, the error bars on a graph claiming to accurately represent the gender proportions of the WikiProject would dominate the graphs. There are many reasons why an editor might choose not to self-identify a gender, and these are discussed in the next section, however the information presented here does provide a window into the topic and may allow some insights into which factors lead to increased female participation.

The data seem to suggest that WikiProjects can be roughly organized into three distinct tiers based on the number of female editors claiming membership. The reasons for this clustering of data into discrete tiers is not obvious, and may not bear out further examination into the topic. Indeed, both WikiProject USA and WikiProject Articles for Creation (neither shown in the graphs) fall somewhere between the low and middle tiers (both between 5% and 10%) and are consequently difficult to place. It is quite possible that analysis of a larger number of WikiProjects would blur the distinction between the low and middle tiers yet further, however it seems apparent that WikiProjects with similar percentages and with similar growth curves are exerting similar effects on editors of declared gender regardless of the tier they are considered to belong to.
 * Low tier - The low tier (ranging from 1% to 5% female membership) includes groups such WikiProjects as as Football, Video games, Math, Deletion sorting, Military History (all found on the "Low scale" graph above at left), and India (not shown).
 * Middle tier - At 13% to 17%, WikiProjects Spam, Film, Albums, Medicine, Biography (on the "Low scale" graph), and Television (not shown), as well as WikiProjects LGBT and Countering Systemic Bias (both in the "High scale" graph) all fall within the middle tier.
 * High tier - WikiProjects Gender Studies, Feminism, and Women (all on the "High scale" graph above at right) span roughly 50% to 70%, and represent the top tier. Note however that caution must be taken before interpreting WP:WOMEN's association with as meaningful. As a new WikiProject, WP:WOMEN is most likely undergoing volatile changes associated with the launch of any new WikiProject (see the first few years of any of the other WikiProjects' growth curves).

Regarding the curves themselves, one may notice that data between 2003 and 2008 (on the "Low scale" graph) appears to be particularly erratic, with most WikiProjects varying wildly between 0% and 15% female participation shortly after their founding. The reason for this is most likely due to the unreliability of very small sample sizes. When a WikiProject starts its first few years with total membership in the single digits or teens, it becomes difficult to regard female editors as a group. The membership decisions of an individual female editor may swing the nascent project's gender percentages in a manner not truly reflective of a gender-related trend, but rather reflective of an individual woman's activities. As such, caution should be taken in regarding the early part of the curve of any of the WikiProjects. It would be folly, for example, to draw conclusions from a contrast between the rates of change in percentage for the earliest years of WikiProjects Math and Video Games versus those of WikiProjects Spam and Gender Studies. Nevertheless, useful information can be gleaned from these curves if they are considered in their entirety.

The general flattening of the great majority of curves in time may be regarded as a more or less accurate reflection of the true levels of gender-declared interest in the projects, and the mild increases in declared female membership in projects like WP:MED, WP:WPSPAM, and WP:LGBT between 2010 and 2012 may also reflect encyclopedia-wide female recruitment efforts overseen by Sue Gardner and the WikiMedia Foundation (WMF). These efforts came in response to earlier survey data collected by the WMF in 2008/9, twice in 2011, and in 2012, suggesting that encyclopedia-wide rates of female editing hovered between 9% (2011 data) and 13% (2008/9 data). The most recent WMF survey (in 2012) found 10% female participation — all figures falling well between our "low" and "middle" tiers.

Perhaps the greatest apparent success story from our "Low scale" graph is that of WikiProject Albums which displays a noteworthy upward growth curve. Until 2012 this WikiProject displayed classical "low tier" curvature but after 2012 it rose dramatically to its present position in the middle tier. Is such a change possible for WikiProject Video games? What would it take for WP:VG to reach 13% female editor membership? In terms of raw numbers it would require 40 female editors or five times the current number of female membership. Is that possible? Well there are currently more than 500 members of WP:VG so if only 1 in 10 members were to actively recruit 1 female editor for the project then we would overshoot our goal by 3%. It's clearly an achievable goal, but is a one-time shot in the arm all that is needed to dismiss WP:VG's Gender Gap? Probably not. Numerous factors based in both systemic/rule structure and in behavior contribute to the apparent editorial gender gap illustrated by the above graphs and one of the most significant of these is the "unknown factor" of the many undeclared editors.

The unknown factor
The three graphs immediately above represent the number of female (green), male (orange), and gender-undeclared (yellow) editors from three different WikiProjects selected to represent each of the three previously identified gender-demographic "tiers". In the left-most graph we see WP:VG — a project that is more than half gender-declared and that is dominated by editors who self-identify as male. In the middle graph we see WP:BIOG — a project that is 2/3 gender-undeclared and that is somewhat less dominated by male editors, and in the right-most graph we see WP:FEM — a project 2/3 gender-undeclared at which males actually for the minority. A question immediately springs to mind: What is the true gender composition of the large undeclared fraction? If we only knew this then we could know for certain where we stood with regard to gender. This question may not be easily answerable.

Wikipedia is an outpost on the internet — a landscape where nobody knows you're a dog. Studies have repeatedly shown that in settings where anonymity is the norm and where gender declaration is not verified, women tend to preferentially make no declaration and may even adopt a male persona in order to avoid various forms of discrimination and harassment. In Nick Yee's article, "Map of Digital Desires" (published in the 2008 Beyond Barbie and Mortal Kombat: New Perspectives on Gender and Gaming), he states that "female players have learned that it is dangerous to reveal your real-life gender in MMOs ... They must either accept the male-subject position silently, or risk constant discrimination and harassment if they reveal that they are female." This should resonate for Wikipedians as it has been well known that Wikipedia is an MMORPG for more than a decade. At the Ninth International Conference on Computer Ethics in May 2013, Robert Bodle pointed out that men have also been found to use the internet pseudonymously and sometimes to adopt female personae for purposes that can go beyond role playing up to and including the harassment of women. Research like this weakens our ability to make as much meaning from the raw gender data as we might hope.

Apart from the purposeful non-declaration of gender, other factors may also lead to the preferential use of the gender-neutral persona. It is worth considering, for example, that self-reporting in a public context where "unknown" is the default option may lead to more "unknowns" than self-reporting in more of a one-on-one context such as email or via anonymous surveys without a "non-gendered" option, where the only way to declare "unknown" is to fail to respond to the email or survey instrument. Also worth considering is that declaration via account settings is a little more technologically advanced than declaration via direct communication. [N.B.: For those who don't already know, gender identification may be easily set or changed via Special:Preferences (under "Internationalisation")]. Finally it's important to recognize that the 3-choice "he, she, they" system effectively acts to lump together all of those who fall outside the gender binary and treat them as a single gender group. Gender queer, non-gendered, and all other extra-binary identities are classified as "they", together with males and females who are interested in obscuring their genders and those that simply don't know how or don't care enough to self-identify as either of the traditional genders. This may have the effect of artificially boosting numbers of undeclared members.

Additional points worth considering include:
 * Membership rolls — This study only encompasses declared/listed members. This neglects to consider the contributions made to the gender demographics by unlisted associated (editors who would never join a club that would have them as a member, for example), as well as those whose membership in a WikiProject is more specialized – leading to listings on membership rolls for sub-projects like taskforces only.
 * Changing genders — A member's declaration of gender may have changed in time. Gender-undeclared editors may have switched to male or female simultaneously with the making public of a real life identity. In other cases, changes in gender declaration may match actual changes in gender (e.g. in gender fluid members), or in anatomy. Member rolls within WP:LGBT, for example, list editors who have declared themselves to be "pre-operative" and later can be seen to have removed that label. These editors would of course be the same person post-operation, but the anatomical changes may have prompted a change in listed gender. [N.B. - Obviously your author is not suggesting that anatomy should be or is the most significant factor in prompting a user to change gender designations]. Editors identifying as male or female may have also switched to "gender neutral" in the interest of greater anonymity or to designate a gender outside of the gender binary. The tools used in this study do not touch on historical gender declaration data. Thus the presentation of historical trends in the above graphs reflects current-day declarations. This may skew modern consideration of the gender climate of WikiProjects from historical periods.
 * A note on the WP:VG graph above — WP:VG (as well as WP:TV) has made use of categories instead of lists as a means of collocating members in the past. Sadly this means that usable membership data for the period between 2006 and 2008 (for WP:VG) and between 2007 and 2014 (for WP:TV) are impossible. For the purposes of the WP:VG graph above, data were extrapolated to follow a linear growth from 2005 to 2009, however this is unlikely to match reality very closely. The last datapoint available for WP:VG membership prior to the adoption of categories comes from mid-2006 and already shows that the "undeclared" fraction listed above for 2006 is too low. These gaps in the data make a strong case for the (non)duplication of lists and categories discussed at WP:NOTDUP.

Analysis
The table above presents the ratio of WikiProject members declaring a gender (either male of female) to members opting not to select a gender (either because they do not fit within the gender binary, because they wish to maintain greater anonymity, or for any of the other number of reasons mentioned earlier). Alongside this ratio, the above table presents the "tier" category identified while examining female to male ratios above. These data seem to indicate that female membership percentages are lowest in WikiProjects for whom gender declaration is more important and highest in WikiProjects for whom gender declaration in less important. One possible explanation might be that undeclared editors are mainly male and that in WikiProjects where declarations are more commonly made, the higher male to female ratio reflects larger numbers of otherwise-undeclared males. Another possibility is that this simply reflects the findings reported by Yee in Beyond Barbie and Mortal Kombat, that women are more likely to be incognito online. Under this line of reasoning, the majority of undeclared editors are women and their accumulation at high-female-percent WikiProjects reflects their interest in common topics.

Curiously, a few outliers including WP:WPMATH (with many gender-undeclared editors and a high male to female ratio) and WP:WOMEN (few gender-undeclared editors and a low male to female ratio) are recognized in the chart. Some amount of explanation can be found from the outset - the fact that WP:WOMEN is as new a WikiProject as it is, suggests that it may yet be in the erratic pre-establishment (i.e. prior to the WikiProject's achievement of recognition and "well-established" status) period where numerous factors make characterization difficult. However from outliers like WP:WPMATH and from interesting or surprising curves such as that of WP:ALBUM in the "Low scale" graph at the very top left, it may be worth looking for structural/systemic explanations for the unusual data. If we can identify structural factors that promote greater gender balance on Wikipedia or that work to widen the gap then we may be able to take steps as an encyclopedia to adopt new measures that might halt or reverse the trend. There is great excitement in some corners of Wikipedia for solutions to the gender gap and if the tone (of candidates and questioners alike) at ACE2015 is any indication, 2016 holds great promise for important inquiries and the further development of efforts in this area.

Conclusions and related readings
How does WikiProject Video games compare to other WikiProjects when it comes to editorial gender disparity? The results of this quick investigation show that WP:VG is not an outlier. While we must recognize that the project does in fact have the second lowest ratios of females to males of all WikiProjects examined, the fact of the matter is that very low ratios like this are relatively common on Wikipedia, and WP:VG finds itself in the less-than-one-in-twenty company of WP:MILHIST, WP:DS, WP:MATH, WP:FOOTY, and WP:INDIA.

In terms of its ratio of gender-declared to gender-undeclared members, WP:VG finds itself to be very much in the mainstream, with a rough one-to-one ratio positioning it alongside most of the same WikiProjects with comparable female to male ratios (MILHIST, DS, FOOTY, and INDIA). Noteworthy is the fact that when considering this ratio, WP:VG is not comparable to WP:WPMATH, and that it does find itself in the added company of WP:USA and WP:WOMEN (with cautions pertaining to WP:WOMEN's youth as a project discussed earlier). In contextualizing WP:VG's declared-to-undeclared ratio with its male-to-female ratio we find the project to fall well within the mainstream as a project with high levels of gender declaration (4th out of all 19 WikiProjects analyzed) and low levels of female editorship (18th out of 19).

An examination of the graph of female participation over time shows that WP:VG is experiencing a mild plateauing at around 3% female editorship. The perceptable bump in female editor membership between 2009 to 2011 is closely mirrored by a spike in overall project size visible in the WP:VG graph of declared to undeclared ratios, and is most likely linked to concomitant adjustments to the member rolls. During this period, inactive members were moved to "Inactive" (a member category falling outside of this study) and blocked/banned members were struck from the membership rolls. The early 2005-era spike in female membership of 7% (more than double the 3% plateau of the last few years) is most likely due to WP:VG's youth as a project more than to its particularly female-friendly practices and the general trend of the graph is most likely a reflection of the natural state of the project without any specific Gender-Gap-related forces at work.

So where does that leave us as a project? We have a large gender gap and we haven't done much to address it. The question which has provoked so much debate and consternation in other areas of Wikipedia throughout the last several years has been whether or not the gender gap is in fact a problem. While those like former Wikimedia foundation executive director Sue Gardner see it as a fundamental problem confronting and limiting the encyclopedia, others including essayist and author Heather Mac Donald have described it as a "non-problem in search of a misguided solution." If the reasons underlying the gender gap were related to blatantly targeted and systematic hostility and harassment then the immediate outcry from WP:VG members would necessitate a hasty intervention and the matter would be quickly resolved, however that is not what we see in actual day-to-day editing. And that fact is important to recognize for outsiders who may only know of WP:VG in relation to Wikipedia's misogyny-centric article on GamerGate. Gamers as a group may not be noted for welcoming women into the subculture, but if your author's experience is anything to go by, the vast majority of WP:VG members are glad for the opportunity to work with fellow enthusiasts regardless of their gender.

To what, then, do we attribute the gap? Has it arisen in response to negative pressures against women more subtle than outright discrimination and harassment? Does the disparity reflect broader societal disparities? Just as WP:VG's gender gap is symptomatic of the larger Wikipedia gender gap, perhaps Wikipedia's gap is symptomatic of societal gaps like the underrepresentation of women in video gaming and women in STEM fields. And are these societal gaps related to intrinsic disinterest? Learned behavior resulting from social pressures? Historical disadvantage? There are indications that the severity of the gap may be somewhat overinflated when self-reporting leads to gender declaration for only about half of editors in most cases and when case studies like those discussed by Nick Yee in "Map of Digital Desires" reveal that women are less apt to reveal their genders online than men. But then is the gender gap nothing more than a statistical fiction? That seems unlikely.

One explanation that may be seeing a slight surge in interest recently relates to Wikipedia's atrophied 4th pillar: Editors should treat each other with respect and civility. Its enforcers not infrequently disparaged as "civility police", the civility policy and its associated guidelines like WP:BITE were established in order to protect good faith editors from seeing their voluntary contributions met with rancor from an entitled set of established or simply aggressive fellow editors. Over the years since its establishment, the 4th pillar has been eroded somewhat with some degree of precedent behind a caveat that perfect civility was not demanded of editors who made up for it with excellent content contributions. More recently, however, Wikipedia has seen increased interest in the concept of civility and this may be a step in the direction of closing the gender gap. One of the common modern criticisms of Wikipedia is that it has become increasingly legalistic, litigious, and prone to policy wonkery. The main effect of this may be to drive away new editors. If the gender ratio of Wikipedia's editing pool is to be brought closer to that of real life while at the same time the encyclopedia strives for editor retention, then an influx of new female-identifying editors is necessary. Perhaps the best way to retain these new editors and to reduce barriers for new editors is simply to redouble our efforts to avoid BITEing.

For readers interested in finding more information about the Gender Gap on Wikipedia, here are a few excellent sources of information:
 * Bayer, Tilman. "How many women edit Wikipedia?". Wikimedia.org. 30 April 2015.
 * Gardner, Sue. "Unlocking the Clubhouse: Five ways to encourage women to edit Wikipedia". Suegardner.org. 14 November 2010.
 * Mac Donald, Heather. "Wikipedia Is Male-Dominated. That Doesn't Mean It's Sexist". Slate. 9 February 2011.
 * WP:CSB's Gender Gap Task Force – Media and Research
 * "" – The Wikipedia Signpost series on the topic.