Research:Men of Quora use Wikipedia to get their knowledge while women have other sources

From Meta, a Wikimedia project coordination wiki
Contact
Laura Hale
This page documents a completed research project.


This is part of a series of research pieces published on Quora Numbers focused on understanding Quora's Gender Gap, and how that community functions. This is published here and mirrored on Quora because of its topical intersections.

Project Summary[edit]

Men are twice as likely to link to Wikipedia in their answers as women, even as both men and women both provide links in their answers at the same percentage.

Introduction[edit]

So this is another week without internet, or rather having only limited internet mostly via my phone. Any research I want to do thus needs to be written after manually pulling from mobile, and then recording that on a separate device.

Anyway, I got asked to answer a question about what percentage of questions contain links to Wikipedia. This is reasonably interesting in terms of understanding how Quora knowledge, when coming from explicitly stated external sources, is formulated through source selection.

My hypothesis going into this is the percentage of answers linking to Wikipedia would be around 0.5% to 5%, with percentage of links in answers at between 1% and 10%, and with some topics having a higher link rate than others because of need for external technical discussion, signatures containing links[3] or issues with link spam.[1] This hypothesis is largely based on the assumption that many questions are formulated around asking opinions, instead of asking for facts.


Methodology[edit]

On my phone, find answers. Record the question, topics, who answered the question, gender of answerer. Also record if the answer links to external content, and then if the link is to Wikipedia.[2] Internal links to other Quora questions or users or answers were not counted as links. Embedded videos and images were also not counted as links, unless there was text sourcing Wikipedia for the image or video. Collapsed answers were included.

The topic plays an important role because dating and relationship questions seem like they will be less likely to require a link to Wikipedia, where as a physics or math question may. The tags used were the topics found on the question.

This is less random than it might be as questions largely were pulled from my own feed because of the constraints at the time of data collection requiring use of my phone. The implications of this are that, among questions selected for inclusion, there may be less representation of culturally Asian Quora because I follow fewer people from this community. This means that this really could do with repeating in the future to see if the numbers generally agree in terms of the low percentages.

The data was gathered on June 18, June 19, and June 20. Data collection stopped at around 1,300 answers as a result of time constraints.

Results[edit]

With 1,328 total answers written by 1,028 unique contributors included in the sample, 10.9% of all answers contained one or more links. Of all answers, 6.8% answers contained a non-Wikipedia link and 4.1% of all answers contained a link to Wikipedia. The most popular topics for Wikipedia links of those sampled were World War II, The United States of America and U.S. Politics.[7]

Gender and Wikipedia linking[edit]

The sample included 1,037 answers written by males[4] , 223 answers written by females,[6] 55 anonymous, and 13 by Quora Users.[5] The specific breakdown for gender and user category type and their prevalence on Quora from this sample is as follows:



Men are twice as likely as women to link to Wikipedia in their answers, even as men and women are about equally likely to provide a link in their answer. This is perhaps one of the most surprising things found in this research, and it is unclear why this pattern exists. According to Alexa, Wikipedia has slightly more females in their audience than the average internet property. At the same time, Wikipedia has a gender gap similar to that of Quora, with around 10 to 20% of English Wikipedia’s editor base being female.

If the total links to Wikipedia are added up, women provided 9.8% of all answers with links to Wikipedia. This more or less matches the low end estimates of female participation on English Wikipedia, but not readership on Wikipedia, nor Quora participation numbers.

Sook Lim & Nahyun Kwon wrote “Gender differences in information behavior concerning Wikipedia, an unorthodox information source?” in 2010 in Library & Information Science Research which focused on gender differences among Midwestern American university students. It found that male students were more likely to use Wikipedia, discount any risks associate with citing Wikipedia, and overall had a higher perception of the quality of information than female students.

Johnny Snyder published “Wikipedia: Librarians’ Perspectives on Its Use as a Reference Source” in 2013 in Reference & User Services Quarterly. The author surveyed librarians in the western United States. He found male librarians were more likely to use Wikipedia than their female counterparts on a more regular basis, at 2.5 times per week compared to 2.0 times per week.

Despite the differences in readership that suggest higher female reading rates from Alexa, these studies could account for why men linked to Wikipedia 40% of the time when linking while women only linked to Wikipedia 20% of the time. Men and women could have different attitudes regarding the reliability and importance of Wikipedia, and this is expressed through linking patterns on Quora.

A better approach to looking at this issue instead of falling back on Wikipedia related studies is to consider knowledge formation. This is important because the type of knowledge formation and the ability to express knowledge differs vastly between these two systems: Quora and Wikipedia.

Sangmi Chai, Sanjukta Das and H. Raghav Rao explored issues in online knowledge sharing in their 2011 paper, “Factors Affecting Bloggers’ Knowledge Sharing: An Investigation Across Gender” that was published in Journal of Management Information Systems. This paper found that offline social norms persisted online.

Sanjib Chowdhury and Megan Lee Endres published “Gender Differences and the Formation of Entrepreneurial Self-Efficacy” in 2005’s Entrepreneurship in Diverse World, volume 8. They found that men exhibit higher levels of self-efficacy across all educational levels, while women with higher educational levels expressed greater levels of self-efficacy. In this context, further research could potentially be done to exam if there are differences in women’s linking to Wikipedia articles based on their level of educational containment.

Additional research needs to be done to look for relationships between gender and knowledge formation to see if potential explanations exist or potential alternative research methodologies relate to knowledge formation as it pertains to Wikipedia can be derived for new research.

Wikipedia Gender Linking and Topics[edit]

Women’s answers with links to Wikipedia came from 20 different topics. Men’s answers with links to Wikipedia came from 96 different topics. The 2 anonymous answers with links to Wikipedia came from 9 different topics. The 2 Quora User answers with links to Wikipedia included 7 different topics.

All three groups were relatively diverse, in that there was not huge overlap in terms of topics being found exclusively for Wikipedia linking. A better way of visualizing this is to see where Wikipedia links are used as a percentage of all answers in that topic.

Women had non-Wikipedia links in 79 different topics, and there were big differences in what answer topics they provided non-Wikipedia links to.

There was some crossover with some topics, but 12 of the 19 topics on the Wikipedia list did not match the other links. This suggests that when women provide links to Wikipedia, they do so because of specific topic needs that may not be met by other sources. It could also be a function of different topics have different types of questions, which make it easier for women to use or not use Wikipedia links.

Conclusion[edit]

This area of research is potentially hugely interesting when it comes to knowledge formation on Quora, and where people get their knowledge or source their knowledge.

Some of the results here, especially the gender ones, are very surprising. The fact that women under use Wikipedia as a source in terms of Wikipedia readership and when compared to men on Quora is not something that can easily be explained. It is also interesting that both men and women on Quora appear to provide links in their answers at a similar rate.

Additional research needs to be done on how knowledge on Quora is formed and shared, and how this can be connected back to other knowledge sources, especially Wikipedia. Both sites share in common a known and documented gender gap. Both are about knowledge sharing, though Quora allows for expert advice, non-neutral points of view, and does not require citation. This sort of editing environment is far different than Wikipedia, because it does not rely on a topic having reached a threshold of importance before being written about, nor reliance on sources that explicitly provide a male dominated perspective.

These Wikipedia weakness should be advantages when it comes to women contributing to Quora, though the site has not been able to capitalize on this where Wikipedia has failed. How are women forming their knowledge and expressing it on Quora? How much of a role does self-efficacy and educational attainment for women play into their overall participation rates compared to their male peers? Is it possible to do research to compare the different levels of educational attainment on Quora when it comes to gender to actually ask that question?

In the Wikipedia world, this issue is important as it suggests that despite available data that suggests men and women read Wikipedia at similar rates, women are much less likely to cite it as a source. This highlights that ongoing efforts to make Wikipedia’s gender gap smaller do not appear to be working as women are not using it and sharing it as a potential knowledge source. If women cannot and do not use it for this, it makes it much more difficult to get this cohort to become actual editors: This lack of trust and sharing of Wikipedia content is yet another barrier to participation.

Quora is just one social media site, and little research appears to have been done using Facebook, LinkedIn, Twitter, LiveJournal and other social media sites to see if similar patterns regarding organic link creation and knowledge sharing through Wikipedia are similarly gendered. If the Wikimedia Foundation is serious about fixing this gender gap, they need to exam external link creation to better understand how people at this time are sharing knowledge, and how this knowledge formation differs from cohort to cohort. Doing so is important for being better able to situate themselves going forward by being able to create better branded messaging to specific audiences.

As the data currently stands, there is room to do secondary research looking less at the links and more about gendered participation by topic going forward. With 1,328 answers, this particular data set is not overly small so applying Social Research Analysis to it may be a viable option to see if new patterns can be observed or other patterns explained base on previous topic based analysis. This research is also not largely a function of time, except that the number of available answers can change, as it is less contingent on dynamic data like follower counts, views, upvotes and answer ranking.

Additional research would ideally work by building on the initial dataset to increase it to around 5,000 to 10,000 answers while working to increase the number of questions and answers about topics more specific to the Indian audience on Quora who are arguably under represented geographically in this sample. The same is also true topically, with questions about sports, sciences, economics, social media and startups under represented.

Including these additional answers, coupling this with some potential survey research, and examining the types of questions included in the research, could better explain how knowledge is formed and shared on Quora. This understanding could have long term benefits for Quora because it could provide a much better framework for sharing with new authors the best ways to share knowledge through sources, and result in an increase in overall quality on Quora. At the same time, it could also give insights into why Quora has its gender gap and how this can be fixed. Fixing Quora’s Gender Gap is important, because women make up a large percentage of untapped Quora users. If they are not tapped in such a way that also continues to attract men, then Quora risks potential stagnation and degradation in answer quality.

Footnotes[edit]

1. Indeed, I found one answer I had reported for link spam dating back to May after I went to report it. The amount of link spam found though was relatively low, at around 1 or 2 links at most.

2. This could result in a situation where someone links to Wikipedia and another source, but the other source is not really seen as a separate link.

3. A few of these were found, and they were counted as external links, but not links to Wikipedia as none of the sig links found were to Wikipedia. The amount of link signatures found though was relatively low, at around 1 or 2 links at most.

4. People who present as male either through real name, picture or self-description found in writing. See earlier research on gender identification methodology on Quora.

5. These are accounts that have been either delete by contributor request, or because they were banned from Quora.

6. This puts female participation at 16.8%, on the low end of suspected female participation rates on Quora though still within the established range.

7. Because of the sampling method used here, India was under represented, both in terms of total questions and in terms total answers. No India questions, topic wise, were included.


Analysis conducted by LauraHale. Raw data used for analysis available upon request.

External links[edit]