Recently there has been a lot of discussion of the suppression of women’s writing in SF with plenty of reactionary response, interesting perspectives, data light discussions, and at least one review website. There are way more conversations going on right now than those few links; I’m simply playing favorites. It’s all interesting and thought provoking, but I’d like to see a conversation that’s a little more grounded in data that can be analyzed for trends. You know, not just in a single data point that’s easy to call bad math (even if the actual math doesn’t change the conclusion drawn).
Earlier this year, Niall Harrison put together some interesting data for 2010 publications and reviews that I don’t think has gotten nearly enough love. I mean, look at that, 43% of the combined US&UK SF books published in 2010 are by women. Why isn’t that brought up by both sides of the debate at every opportunity? It’s a delightful little romp, but it leaves me with many questions. How typical was this year’s distribution of published authors? He’s linked to some older statistics, but they’re missing a lot of years. Also, while I’m fascinated by the apparent trend in what is being reviewed, what is actually being read?
If only there were some sort of popular award that had been running for at least 50 years that we could look at. Something that might be useful for trending what is being read to see if the gap is changing. Oh, right, there’s the Hugo Awards!
Someone else has even taken the time to total the four fiction categories by gender. I decided to look at just the novel category. I suspect that is what the average person is thinking of when they think of a Hugo. Conveniently, Wikipedia has a list of all nominees back through 1959 and even links to descriptions of most of the authors so I don’t have to guess at gender. That’s right, people, I am using wiki-fucking-pedia as a source.
The 2011 ballot is out, but at the time of this writing, voting is still open. In a year where 43% of novels published were by women, 4 out of 5 nominees in the novel category are women. Before we break out the Champagne or lament the suppression of men, I should also point out that 3 of the last 5 Hugo Awards for Best Novel were awarded based on ballots without women. The 90’s only had one year with an all male ballot. This century, we’ve had 4. That means 40% of the Hugos awarded this century didn’t shortlist a woman for the novel category.
Let’s start with some easy math. Between 1959 and 2010:
- 18% of nominees and 27% of winners have been women. Actually, of the 15 female winners, 4 have been Lois McMaster Bujold.
- Only 9 individual women have won a Hugo for best novel compared to 28 individual men. That’s 24% of unique winners.
- 21 ballots have had no female author
- 2 ballots have had more women than men (1979 and 1992).
To look at nomination trends, I decided to plot the percentage of women on the ballot by year to see if it looked like it was improving and got a wide scatter of blue. Then I asked someone who is much better at statistics to look at it. He suggested that if I squint hard enough, I might be able to find some trends in what looks like random noise by examining moving averages and frequency distributions. I went with a 10 year moving average, plotted in red. That is, points on the red line are derived by averaging that year’s ballot distribution with the 9 previous years so that instead of a snapshot, the number represents a 10 year trend. I also included a nice grey reference line for 50%.
So, looking at the red line, women’s representation was improving until the turn of the century. This century, not, say, Mary Shelley’s time.
How about the frequency of all male ballots?
(Double click for larger image)
1963 is the first year a woman was on the ballot. Seven years later, a woman won for the first time. Again, the trend looks good until this century, where the curve actually starts to rise ever so slightly.
Next, I wanted to see how women actually did when they managed to get nominated. From the rough numbers bullet pointed above, they appear to win more often than anticipated based on their representation. That is, all things being equal, if 18% of the marbles in the bag are one color, only 18% of the marbles pulled from the bag should be that color. Not 27%. It is a rough number, though. Instead I decided to look at the gender distributions of individual ballots to see how women did.
First, I needed a comparison point, so I built a table showing what the results of a random draw would be for various distributions of gender. It’s a table and not a list because ballots have had between 4 and 6 nominations total. Note, I went as far as five women on the ballot which has never happened. It has happened with men (so has 6) and their random probability of winning table looks the same.
|Number of Women on the Ballot||Random Probability with 4 nominees||Random Probability with 5 nominees||Random Probability with 6 nominees|
Using the data through 2010, the likelihood of winning based on the gender makeup of past ballots can be compared to the statistical likelihood of winning at a random draw. For years with more than one winner, both winners count towards the gender total while the ballot only counts once towards the total number of occurrences.
- If exactly one woman is nominated, a woman wins 41% of the time.
- If exactly two women are nominated, a woman wins 50% of the time.
- If exactly three women are nominated, a woman wins 100% of the time.
- If exactly four women are nominated, a woman wins 100% of the time.
Guess what? When 3 or 4 women are nominated, they always win. 100% of the time. Bad news for Ian McDonald this year. That’s right, all two times women have been the majority . . . well, I guess that doesn’t really say a damned thing, does it?
Those first two cases, though, deserve a closer look. Of the 17 times only one woman has been on the ballot, she’s won 41% of the time. That’s considerably higher than the 17-25% prediction. With two women on the ballot, the results are closer to the prediction, but still slightly on the high side. So excluding the two female majority ballots, women still do better than a random draw when they can actually get onto the ballot.
What’s the take away from this analysis? Strategies for winning an award? I suppose that if you want to win a Hugo and are a woman, you want to be the only woman on the ballot. If you’re a man, you don’t want any women on the ballot. As those are the two most frequent distributions, probability is on your side.
Joking aside, all I am showing is that the conversation is still relevant. The last decade has seen not just the continuation of fewer novels by women being nominated for Hugo Awards. There’s actually been a demonstrable decline. Why is that? Why has looking at more data only left me with more questions?
We readers are the end of a long chain that at any point can inject bias. Is it a case of fewer women writing, getting published, getting talked about, or being read? Is it, systemically, a little of all those things? I’d love to get my hands on equivalent data for the last 20 years worth of publications to see if 2010 is typical and on trend on the publisher’s side but lacking on the reviewer/reader side. I’m not really looking forward to having to actually pull together such a data set. Anyone happen to have one on hand? The stats page I linked to above is by fiscal year and not calendar year, which is the basis for Hugo eligibility so even the few points provided there don’t allow for a useful comparison. ISFDB, maybe? I haven’t even touched on subgenres.
Oh, the effort I can pour into this.
In the mean time, I’ve just picked up Blackout by Connie Willis. She’s been on the “to read” list for longer than I would like to admit.
DMS is an over-educated engineer living and working in the Deep South. In her free time, she reads, plays video games, and travels. She wholeheartedly approves of getting into blue boxes with strange men.