Our very own Elizabeth Sampat asked a question the other day on Twitter. “Are female indie RPG designers,” she wondered, “less likely to publish a second game than male indie RPG designers?” She pointed out that several women she knew had very negative reactions to their first games, and this might be affecting their ongoing publication rate.
As a researcher, I immediately wondered whether there was a good way to answer the question in a more general way. I knew I’d have several limitations; I didn’t have access to my office computer last week, with its powerful statistical software, and I didn’t have much time to spare. Even given those constraints, though, I was able to get a reasonable answer pretty quickly. Here’s how!
Step one: choose a research strategy.
I had to decide what I’d count as a successful answer to this question. That would tell me what data to collect and what to do with it.
Male and female RPG authors may have qualitatively different experiences in the publication and review process, but a question about percentages told me I’d be counting authors of published games rather than, say, interviewing them. With a large enough game sample, I could use statistical analysis to compare male and female authorship rates.
Once I had the numbers analyzed, I could decide whether to follow up with more of the same, or whether to shift to a different strategy.
Step two: choose a data source.
I knew I’d need a list of published RPGs to work from – and it had to include author names. The most complete list I found was John Kim’s RPG Encyclopedia. It doesn’t claim to be exhaustive, but it has enough authority to be a believable sample. I also had to chose which data from his site I’d work with. He breaks games down by year; I chose to begin my analysis in 2002 because it was the year of the first Indie Game Awards, and to end with 2012.
Unfortunately, the data was all HTML-formatted. For processing and analysis, I wanted a consistently formatted CSV file. (A CSV file separates each piece of information with commas; it can be read by most spreadsheet and statistics programs.) I wrote a perl script that took the data and converted it for me. For each game Kim listed, I grabbed the name of the game, the year it was published, the publisher, and the names of each author. Now I had something to work with!
Step three: process the data.
I needed to do three things in my data processing: identify indie games, identify author gender, and count up the number of games for each author.
I almost immediately realized I didn’t have a good way of identifying indie games. I had over 750 games in my sample; I didn’t have the time to personally investigate each one. Rather than get hung up on this step, I decided to just look at RPGs as a whole. I could always come back to it later!
Next, I realized I could count and identify gender at the same time. I used perl to create a list of all the authors in my sample – a total of 886 authors with 1358 different mentions. I imported it into Google Docs and sorted it alphabetically. That meant that if a person was listed as an author for multiple books, I’d see their name repeated several times. For example, Emily Care Boss was listed six times in a row because her name appeared on six different games in my original data.
I created three new columns in my spreadsheet: “Count – Male,” “Count – Female,” and “Count – Unsure.” I counted up the number of times the name appeared and put it in the appropriate column. That meant Emily Care Boss had a 6 in the “Count – Female” column.
I knew this had some flaws. It couldn’t account for women publishing under male names, or vice versa. It completely ignored authors who published under their initials, or whose gender I could not easily identify from their names. As with the indie issue, I figured I could take a quick-and-dirty pass and come back to refine the data later. For the moment, I had usable numbers on over 700 authors. That would be plenty for a first cut.
Want to know what I found out? Check back for the second half of this article!