Analyze male/female attendance over time

[EliAndrewC]

MAGFest doesn't ask for gender (which is an ambiguous and poorly defined concept, etc) when people register, but we could make reasonable inferences about most people's gender based on their first names. This wouldn't give us an exact percentage, but it would tell us generally what percentage of MAGFest attendees are boys/girls. So we could do the following:

• find a source of data for identifying gender based on names
• take our data for the past few years and see what percentage of MAGFest is male/female and how that's changed over the years

This might be interesting to combine with the data about what percentage of MAGFest attendees return to MAGFest vs just being there once. Are women more likely to come once and never return than men? Etc.

[ghost]

Link for Male Names - https://gist.github.com/JoshMackin/9d1cf9095aac9e5ab462 Link for Female Names - https://gist.github.com/JoshMackin/d1c634c3fa9b0e6ce079 Link for Unisex Names - https://gist.github.com/JoshMackin/3a0a4670949aa43874f5

[kitsuta]

By the way, what's your source for these names?

[ghost]

Here's the links to the site where I got the files:

http://www.infochimps.com/datasets/word-list-3800-common-male-given-names-english-speaking-countrie http://www.infochimps.com/datasets/word-list-4900-common-female-given-names-english-speaking-countr

then I organized them with a little java program I wrote (which is where the unisex name list comes from).

[thaeli]

As a gender non conforming person - please do not do this. I realize you mean well, but the concept is inherently problematic and erasing for those of us with non-binary identities. If we have a business reason to explore this, let's look at what the business need is and work backwards from there.

A set of blog posts I particularly recommend, as they explain this very well: http://www.sarahmei.com/blog/2010/11/26/disalienation/ http://www.sarahdopp.com/blog/2010/gender-is-a-text-field-diaspora-backstory-and-context/ http://www.sarahdopp.com/blog/2008/genders-and-drop-down-menus/ http://www.sarahdopp.com/blog/2010/designing-a-better-drop-down-menu-for-gender/

[EliAndrewC]

To be clear: we definitely will never ask for someone's gender when they preregister (or at any other point). That's not what this ticket is at all, nor will any other ticket ever cover that. The point of this ticket is to try to make some reasonable guesses about what the gender percentages of MAGFest are, and analyze some trends based on that.

The motivation is that MAGFest has historically been extremely male-dominated, and male-dominated conventions are often unwelcoming to anyone who's not a hetero-normative male. If people whose names correlate with being female are significantly less likely to return to MAGFest than people whose names correlate with being male, then I'd really like to know that. Ditto for other similar questions.

[thaeli]

Closing as part of post-M13 cleanup. I still think this is out of scope, and it's not something anyone was getting to anytime soon anyway.

[EliAndrewC]

I'd actually like to keep this open. Although you're probably right that this is something we won't get to soon (unless a newcomer shows up with an interest in data stuff), I'd like to keep a record of it as an outstanding TODO item.

[bds002]

Eli - I'm with Thaeli on closing this as an outstanding issue.. I don't think we have the time or the bandwidth to track things based on first names. Gender ID for names changes over time, as well as personal identities. Even if you have data as per your comment on 16 Sept, I'm not sure what we would do with it??

If someone isn't coming back to MAGFest because they feel unwelcomed then either 1. Our Code of Conduct is out of date (v1.5, last updated 6/17/2013) or 2. the person making the attendee feel unwelcome has violated CoC and we should be addressing that issue on site, or as soon as we are made aware of it.

[kitsuta]

We have a couple of data analysis tickets that I thought I'd get around to, but then it turned out that there's too much stuff to do for the system itself. I think both this use-case and the other ticket would be better-served by third-party reporting software.

That being said, data analysis is very, very tricky. I'd be reluctant to dig into it without someone who's really good at it - it's too easy to draw the wrong conclusions based on data or to make bad assumptions while sorting the data.

In this case in particular, I think our data is just too naive. What exactly would you do if it turned out that attendees with typically-female names don't come back after their first year? You'd have to make assumptions about why they don't, then act on them. If your assumptions are wrong, you risk making things actively worse - diversity initiatives in particular tend to be a double-edged sword.

I think to solve the overall concern here, you want to collect better data. We had a good try with the survey sent out this year, but the survey itself had design issues. Instead, we'd probably have better luck with a simple survey that enters you into a raffle for a free ticket. It'd have something like these questions:

1. On a scale of 1 to 10, how likely are you to come back to MAGFest?
2. Why? (optional)
3. On a scale of 1 to 10, how likely are you to recommend MAGFest to your friends?
4. Why? (optional)

[EliAndrewC]

I'd still really like to see this analysis done. I agree that we shouldn't draw too many conclusions from it necessarily, but if there's an easy way to get a general sense of whether or not women are significantly more or less likely to come back to MAGFest for multiple years than men, I'd really like to know.

With that being said, I have no objection to closing this issue on the basis that Github isn't a great place to track it, and we can always link someone to the closed issue later on if we get a volunteer who wants to dive into this.

[kitsuta]

Now that we're using JIRA for non-code things, this should probably be moved there. I did stumble upon this API, which should simplify a lot of the work: https://genderize.io/

[binary1230]

indeed we should! https://jira.magfest.net/browse/INC-173

[DataFighter]

The key is collecting the right kind of data and you'd need a lot of it over several dimensions. You would have to look at other factors such as age, demographics, distance, events attended, time spent at the con, hotel logged into, etc. etc. etc.

Even if you can prove any kind of correlation, finding a causitive link would be difficult and would require social experiments.

Data science is nice to optimize aspects of the con experience but I don't know if specifically looking at gender is one of the things you should be focusing on. It should be stated, that the gender issue is systemic in gaming, so it could be that the male/female ratio is purely because this is a gaming con (that in itself could be a research project).

One thing I'd suggest would be to try to track attendees behaviours beyond attendence. E.G. What events do they go to ? etc. etc. If you are worried about privacy and such, you could make it opt-in and gamify it. And be honest about what you are doing. E.G. 'We are trying to improve the magfest experience. Help us by participating in game ~~~~ at any events you attend.'