HUGO AWARD-WINNING SCIENCE FICTION & FANTASY MAGAZINE
The Issue of Gender in Genre Fiction:
Conclusions
What is the situation with women in SFWA-qualifying science fiction short story markets? What proportion of publications are authored by women? What proportion of submissions are authored by women? Is science fiction significantly different from the other genres in these markets? Can we see any trends that might explain the differences between markets?
These were the questions I wanted to ask when I began emailing editors back in the early months of 2014. In all honesty, I was not aware of the size or complexity of the task I was taking on when I began this study. As the data began to come in, and I saw the numbers of submissions that editors were categorizing for me, I gained an even greater appreciation for their assistance. This study has, in many ways, been a community project, with slush readers and editorial staff taking time out of their already busy working days to help provide valuable information to the science fiction and fantasy community.
Overall, authors who are women are less well represented in terms of submissions and publications than authors who are men. While some markets published more women than men in both all genres and in science fiction specifically, no market received more submissions from women than from men. However, markets displayed significant differences in terms of both submissions received, and stories published, for both science fiction alone, and all genres.
There are differences between markets, so there is something that makes a difference. What that is, we cannot say from our data. We found some relationships and correlations, but nothing we can point to and say, “This is it. If all markets made this change, we would see a greater representation of women authors.”
Upon deeper analysis of multi-genre markets, I found that there was a correlation between a high proportion of women authors overall, and a high proportion of women authors of science fiction stories. However, we still found that women were, in many cases, not as well-represented in the science fiction section as they were in the market taken all together. Clearly, many of these markets are very friendly to authors who are women, and in some cases are attracting relatively high numbers of submissions by women, and still, science fiction is lagging behind other genres. The cause of this is unclear from our data, but it is an interesting result.
Submissions
Age of senior editorial team, gender of senior editorial team, or whether a market is multi-genre or science fiction only had no relationship with the proportion of submissions received from women.
Publications
Gender of senior editorial team had a moderate relationship with the proportion of published stories by authors who are men, but no relationship with science fiction stories.
Age of senior editorial team had no impact on the proportion of women and men selected for publication.
Science fiction only markets were more likely to publish stories by men than were the science fiction sections of mixed-genre markets.
Given the limitations of the data, it would be misleading to statistically analyze submission data and publication data together. The reasoning for this is given in the previous article, but in essence, it’s because there are differing and inconsistent time lags between receipt of submissions and selection of stories for publication, and between acceptance of stories and the final publication date. We can’t know from our data what the proportions of submissions were for a particular issue’s publications.
No editor considered their submissions data to have an unusual proportion of men and women authors. A few points of percentage difference might not be noticed by an editor, but could lead to significant differences in statistical results. Hence, we can’t really run correlative tests and so on for various factors. That said, we can draw general inferences from the submissions data, when considered with the publication data. Large swings of 10%+ would likely have been noticed by editors, so it seems fair to say that we can see large differences in the relative proportions of men and women selected from slush by different markets.
As you can see from the chart, it is not as simple an issue as saying, “We need to increase submissions by women,” or “Markets who publish more women have greater numbers of submissions to choose from.” While some markets closely track publications percentages to submissions percentages, other markets do not, showing that submissions ratios are not the only factor affecting publications ratios. Basically, we don’t have an easy linear relationship between proportions of submissions by women and proportions of publications by women.
Whether or not increasing submissions from women would lead to an increase in acceptances from authors who are women is not a question I can answer from the data. I would need to compare markets over time, controlling for other factors that might affect acceptances. In addition, due to the time-lags and variation between submissions and publications, I can’t run the tests necessary to say there is a statistically significant over-representation of women in publications, as compared to submissions. I can say that from a general reading, markets are at least matching their proportions of submissions, with some individual markets publishing a greater proportion of authors who are women than they have in submissions.
One useful outcome from this study has been a decision by some markets to update their guidelines to explicitly welcome submissions from authors who are women. Should this increase submissions from women, the time-study mentioned earlier would be possible, and the question of the relationship between submissions and publications ratios might be more easily answered.
At the start of June, Escape Pod made use of this strategy. In terms of raw numbers, their June submissions from authors who are women increased from an average of 19 of 75 total submissions (25% share), to 27 of 83 total, making up 32.53% of that month's submissions. July (up to the 29th of the month) showed a similar increase, with 36 submissions by women of 111 total, for a 32.43% share of submissions.
This is only a single market, and two months of data. However, apart from the updated guidelines and the announcements of such on Twitter, the only other changing factor, according to the staff, was an increase in pay to the new SFWA qualifying rate. While it is possible that this increase led to a greater number of submissions, it would not explain why women were more motivated by that increase than men. There is always the possibility that a third, more unlikely factor has caused this increase, but it does give an indication that such explicit welcoming of women may increase their share of submissions.
According to Nathaniel Lee of Escape Pod, their reasoning was simple. They discovered that they were receiving only a small proportion of submissions from women, and felt that the easiest-to-implement first step would simply be to encourage women to consider Escape Pod as a market. Whether this increase is a transient response or a sustained development is currently unclear, as is whether it will make an impact on the proportion of published stories by women. Even still, either result will give fruitful data for analysis and discussion.
One area in which this study was lacking was in data on non-binary individuals. In most cases, the gender categorization for each submission was based on first name, although some editors also engaged in Googling public bios of authors. As such, there are undoubtedly submissions from non-binary individuals which were miscategorized. Rose Lemberg raised this issue on Twitter, where it generated a fruitful discussion. Excitingly, a proposal was put forward suggesting that markets include an optional post-submission survey allowing for collection of demographic information based on the self-identity of authors. Crossed Genres expressed an interest in such a survey, and if there is follow through from markets, the full picture for both gender identity and other demographic areas would be much easier to analyze.
With that said the fact that we are looking at data on ‘apparent’ gender, does not invalidate the study. Rather, it gives us another factor to consider when we assess the rigor and implications of the results. I am grateful to all those who took part in the Twitter discussion for elaborating on these implications and effects.
My aim in carrying out this study was to answer some of the underlying questions about representation of men and women in science fiction. Discussion of this topic often reaches an impasse when differing contributors disagree on the potential causes of a problem or whether there is in fact a problem at all. I hope that at least some of these issues can now be more easily discussed, with common reference points for the actual circumstances.
However, to say that all the questions are answered is, of course, blatantly false. Firstly, there are the limitations of the data collection and analysis, which I hope have been made clear in all instalments of the study. Secondly, there is the fact that complete investigation of many of these questions would require different data, samples, and methodologies. Thirdly, there is the fact that there are important questions that I have not even attempted to answer.
During my communications with editors, a number of questions were raised which this data could not answer, but which would be fruitful areas for further research and discussion. Are women less likely to resubmit to a market after receiving a rejection? Are women less likely to submit the same story to a different market after it has been rejected elsewhere? Are women less likely to self-categorize a story as science fiction when submitting to multi-genre markets? Are women less likely to submit a story to a market that only publishes science fiction, perhaps due to this self-categorization issue?
I would very much welcome further research aimed at answering these questions. Post-submission surveys, along with the continued tracking of data which many publications have expressed an intention to carry out, will hopefully allow for follow-up and comparative studies. This data will hopefully better elucidate how various initiatives and strategies affect the submissions and publications gender ratios in the genre.
Overall, this study has been time-consuming, occasionally frustrating, and always challenging, yet has, I believe, revealed important data for the community. We have no smoking gun for the differing levels of women’s submissions and publication rates across different markets. And yet, those differences do exist. This is an important realization. Again, I wish to thank all the editors and editorial staff who assisted, and the insightful comments from readers. The engagement and support of the community in investigating gender parity has been immense—we have not yet reached our destination, but our feet are firmly on the road.
ABOUT THE AUTHOR
Susan E. Connolly's short fiction and non-fiction have appeared in Strange Horizons, Daily Science Fiction, The Center For Digital Ethics and the fanzine Journey Planet. She is the author of Damsel, a middle-grade fantasy from Mercier Press and Granuaile, an upcoming historical comic book from Atomic Diner. Her degree in Veterinary Medicine given her strong opinions about the accurate portrayal of animal sidekicks in fiction. Susan lives in Ireland, near the mountains. Also near the sea. Also near the forest (Ireland is a small country).
WEBSITE
Also by this Author
PURCHASE THIS ISSUE:
CLARKESWORLD ESUBSCRIPTIONS AVAILABLE AT
Follow Us On
ISSN 1937-7843 Clarkesworld Magazine © 2006-2015 Wyrm Publishing. Robot illustration by Serj Iulian.
Kevin P wrote on August 4th, 2014 at 6:31 am:
I've enjoyed this series, but I can't help thinking that you're trying to avoid addressing the obvious conclusion. Whether you think it's valid or not, here are the two main conclusions from comparing submission and publication numbers:
1. Submissions by women are 51% more likely to be published than submissions by men. [Source: Chart B1 - men appear to make up 71% of submissions but only 62% of publications, while women make up 28% of submissions and 37% of publications]. I'm looking at SF only here, because that's the main subject, but the ratio is even more unbalanced when looking at all submissions/publications.
2. Of the twelve SF publishing venues, eleven have a higher publication rate for female authors, while one appears to have approximately equal publication rates between men and women. I think it's rather misleading to describe this as "some individual markets publishing a greater proportion of authors who are women than they have in submissions". Even if AE turns out to have a slightly higher proportion of male publications than submissions, that's still a p=0.003 finding for the hypothesis that "submissions by women are more likely to be accepted" [Source: Chart B2b - the null hypothesis is "submissions by men and by women are equally likely to be accepted", and I've used a binomial distribution]
(Yes, I did read the part where you said "it would be misleading to statistically analyze submission data and publication data together", but I don't accept that statement. The statistics work exactly the same way even if you assume that publications and submissions are two independent populations. The difference between male and female submission rates is big enough that any random variation between the two populations would have to be big enough to also rule out pretty much all your other conclusions. And the only systematic variation over time I can think of (increasing engagement by women in the SF field) is far too slow to have a significant effect on the data over the average period between submission and publication, considering that the success ratio is so unbalanced.)
Susan Connolly wrote on August 5th, 2014 at 11:26 am:
Kevin, you've obviously spent a lot of time looking at the data, and I'm glad you're enjoying the series. I did not intend to avoid any conclusions or mislead anyone with the data - indeed, I think that this issue is due to my opposite intent. One of the things I decided was of primary importance when analysing this data was to use the most rigorous statistics that I possibly could - running only the tests that the data allowed, and not making leaps of logic or conclusions. I feel that it's important that people reading these articles could be certain that what I said had evidence or did not have evidence was trustworthy. Note that I am not saying that your conclusions are wrong, simply that running a cross-market test for significance rather than raw percentages was not something I felt the data supported, and without a test for significance we are in a non-specific place regarding the meaning of the percentages.
I very much welcome further analysis and data collection in this area, and hopefully such work will put that question to rest.
Tom Mannly wrote on August 5th, 2014 at 7:55 pm:
I have to agree with Kevin, it was fairly obvious that most publications were publishing with a bias towards women. The fact that you didn't address this is statistical elephant is disappointing.
Neil Clarke wrote on August 5th, 2014 at 8:55 pm:
You're reading the data the way you want to see it. You could also look at the number of published stories and say that most publications had a bias against women (can't have it both ways)... or back at the slush and say female authors are writing better stories than their male counterparts (people have tried). You could keep going down any and all of these flawed rabbit holes, but does the data statistically support your interpretation? That's what makes it true, false, or just a theory.
A big problem here is that quantity does not map evenly to quality. We receive 800-900 stories per month, but the number (not percentage) of "good for us" stories we're receiving is at the same level it was when we were receiving 400-500 stories per month.
One of the other things to keep in mind is that the numbers here are gender by story, not the number of authors submitting or published. I wonder what happens when it's a strict headcount. We don't have data on whether men or women are more prolific here. Could be relevant or not.
If you look at most magazines, they all have "regulars" who can sell them one or more stories in a year. What happens when you remove them? I suppose, it might have been better to look at spots taken by new (to the publication) writers. The stable of regulars in the field has to be large enough to be impacting the percentages by market.
In the end, this is just a starting point for further research. There are more questions generated by the data than the data can provide answers for and asking questions is GOOD. That said, more data is necessary.
Kevin P wrote on August 6th, 2014 at 1:07 am:
@ Susan - thanks for your reply! My initial comment might have given the wrong impression, but I'm not accusing you of deliberately trying to mislead people; I just disagree about whether the difference in male/female acceptance rates can be supported statistically.
I think the main point of disagreement is still the question "can the submission and acceptance rates be compared against each other?". Perhaps another way to look at this is to treat the submission figures from Chart B1 as something constant - if 28% of the submissions in your sample came from women in your sample, we can say that 28% of all submissions come from women, including in the batch of submissions that the publication data came from. Then we can ask the question: given this submission rate, what is the chance that 37% of published stories were written by women?
For the purposes of this calculation, it simplifies things to look only at male and female authors, so we get a submission rate of 28.3% female and a publication rate of 37.4% female. Looking at Chart A3, the overall sample included a total of around 320 publications from slush. So we can use the binomial distribution again to find the probability that at least (320 * 37.4%) of these 320 publications were from female authors. Plugging in the numbers, the probability of this happening by random chance is 0.022% (p=0.002), if the actual rate of publication is equal.
@ Neil: I absolutely agree that the numbers don't prove anything about publisher bias. But I think that they do show that stories written by women are more likely to be published, and that the evidence for that is at a statistically significant level. There are many possible reasons for that (my bet is that it comes down to men generally being more confident/arrogant, possibly added to a bit of perceived anti-female gender bias that makes women less likely to submit stories unless they're really sure of their quality), and understanding these reasons can only be a good thing for everyone involved.
If I seem negative, that's not intentional at all - I've found this series of articles really interesting, and more data can only be a good thing. I just have a bad habit of writing about things I disagree with and not bothering to mention everything I like.
Susan Connolly wrote on August 6th, 2014 at 10:13 am:
Kevin, no worries! I think you're correct on where our disagreement lies. In the second article, I noted that taking a sample as the constant rate of submissions is problematic due to variances from month to month, with examples from Clarkesworld and Escape Pod showing variations of 3.5% and 4% respectively, between the months with the highest and lowest representation of women. The variations for other markets could well have been smaller or larger, depending on the sample. Sure, editors felt the samples were 'not unusual' but that's not really a statistically rigorous assessment!
As such, running an analysis comparing subs/acceptances for markets as a whole was not something I was confident in doing. It's possible that with some further month-by-month submissions breakdowns from markets we could come up with a 'worst-case-scenario' variance, in which case I would have been happy to run those numbers based on that, but two markets was a bit too small for my taste.
However, I really appreciate your work and I'm sure that many people won't have my persnicketiness about it!
Kevin P wrote on August 7th, 2014 at 12:08 pm:
Oops, it looks like I made a mistake in my post above. The probability should have been 0.22% not 0.022%. Doesn't change anything qualitatively (and the p value was correct), but a bad mistake to make given that that number was the whole point of my post.
Tom Mannly wrote on August 9th, 2014 at 8:32 pm:
Ever since the sexism scandal last year, all the sfwa publications have been falling over themselves to publish women. They all desperately want to be seen as equal opportunity publications. Some even promoting 'female only' anthologies and issues. This would explain the higher success rate for female submissions. And while I'm all for gender equality in science fiction, I don't think that kind of bias is healthy for the genre. I would love to see the stats before the whole sexism thing exploded.
Neil Clarke wrote on August 10th, 2014 at 7:44 pm:
I know some people would like to believe that, but it simply isn't true. Your position supposes that editors lowered their standards to achieve some political agenda pushed by any organization that has no power over editors based on a controversy that had nothing to do with the fiction qualifying markets published.
Tom Mannly wrote on August 10th, 2014 at 8:58 pm:
To ignore the impact that the controversy has had is silly. This whole article would probably never have been written if the issue of sexism had never arisen. Women destroy science (lightspeed), women and aliens month (drabblecast) and even female only e-zines.
And to say the backlash to the controversy has been vehement is understating it. The lines were very quickly drawn and anyone who spoke out against the status quo was quickly ostracized. So of course it was in editors best interests to appear female friendly. I mean lets call a spade a spade here.
Neil Clarke wrote on August 11th, 2014 at 3:17 am:
And you are ignoring the years of women only issues and anthologies that came in the decade before the controversy. I've also run slush pile statistics editorials several times in the past. It's a bit of a jump to say that sexist images and language in one publication impacted the buying patterns of fiction editors in so many magazines while ignoring the history before it.
Yes, this article was inspired by wdsf, but purchased because we thought real data was important. We would have run the same piece if pitched years earlier.
J. Nelson Leith wrote on August 16th, 2014 at 7:49 pm:
The argument that women's submissions were arguably better than men's is a fair, reasonable, and even compelling argument to make in light of the clear trend (I avoid "bias" for that very reason) toward publishing women's submissions at a higher rate than men's. Sometimes certain demographic communities are in the right place psychologically to tell the best tales. It's a natural part of the ongoing intertextual, intercultural dynamic.
The truly important observation is that a similar but pro-male trend would be almost certainly met with a torrent of knee-jerk allegations of misogyny, especially if anyone dared breathe the exact same "they're putting forth better stories" justification. This article would be making the enraged, viral rounds at Jezebel, Tumblr, Upworthy, and HuffPost by now.
In other words, no clear conclusion can be made about publishing (SF or otherwise) from these raw numbers, but a counter-narrative conclusion about our highly politicized culture is strongly suggested from the lack of political outrage about the numbers.
The irony is that a common prejudice toward seeing sexism only when one sex is on the losing end is actually an expression of bias, not a confrontation of bias.