Following on from Part I, which presented my results, I’m going to discuss in more detail some of my methodology. Yes, that order is backwards. Tough.
If you’re going to find this boring, I suggest skipping ahead to Part III.
Obviously the way in which I select and acquire the books I read will affect my results; that’s the whole point, after all, of encouraging people to seek out diverse authors and diverse representation. Because even assuming an entirely unbiased reader (and which of us is truly entirely unbiased?) there are any number of systemic factors which affect which books get seen, talked about, and ultimately read. Editorial selection, marketing budgets, distributor preferences, bookstore shelving, selection for review, selection for awards…these and more can and do differentially affect female authors, nonwhite authors, LGBTQ+ authors.
So how do I choose my books, and where do I get them? I find books in a number of ways:
- New releases from favorite authors
- Recommendations from friends, acquaintances, authors I follow on twitter, etc.
- Amazon “also boughts” and Goodreads recommendations
- Bookbub emails (which promote discounted or free ebooks)
- Online recommended lists from sources I trust (Barnes and Noble, Tor.com)
- Lists of upcoming releases (I browse the descriptions for things that look interesting)
- Library browsing
- Bookstore browsing
I don’t have hard numbers on which of the books I read last year fall into which category. It would be tedious to go through and classify them, and I probably don’t remember anyway. I can tell you that I got 62 books (roughly 56%) of my books from the library last year. As I mentioned in the last post, this is consistent with my most common year of publication being 2015 — I’m usually aware of stuff as it comes out, but there’s a delay between publication and a book being commonly available at the library.
Anyway, I have lots of sources of books and a lot of the people I’d be culling recommendations from are also invested in reading widely and diversely, so I should be able to find a good spread of books. That said, I’m not immune from the aforementioned systematic effects, and do need to be mindful to deliberately counteract them. I do make a deliberate effort to choose books by women, and I greatly prefer reading about female protagonists. When I say that I deliberately choose female authors, I don’t mean that I read “bad books” just because they’re written by women, nor that I prevent myself from reading books that look appealing just because they are written by men. I do, however, take author gender into account when deciding what books to place on hold at the library, privileging books I am interested in which also have female authors — and a book by a male author has to work a bit harder to keep my attention when casually browsing. I haven’t, however, made any particularly effort to seek out books by nonwhite authors, nor have I, until recently, deliberately sought out LGBTQ+ authors or characters.
Let’s talk about how I did this project. The bulk of the data was exported from Goodreads, where I record and review all the fiction and nonfiction that I read. Data exported from Goodreads included title, author name, book length, ratings, and date of publication. I then went through and (from memory) manually recorded the demographics of protagonists. Author demographics were added by memory or from online author bios and photos (more details shortly). All calculations were preformed in Google Sheets (Excel, basically), and charts were also generated in Sheets.
I classified authors as either male or female based on their names, author photos, and the pronouns used in their online author bios. As far as I am aware, none of the authors included in this dataset identify as any kind of nonbinary gender identity, but I recognize that I could be mistaken about that, or it may not be something that they share publicly. However, I can only work with the information available to me.
While ideally one should strive for true diversity of authors, including all sorts of people (Latinx, black, Native American, other indigenous peoples, Asian people, etc. etc.), I suspected that my reading was not sufficiently diverse enough yet to break down into narrow categories. Just evaluating the percentage of “nonwhite” authors would give me a starting indication of my (lack of) diversity. In addition, information on an author’s race is often not explicitly stated and must therefore be assumed — it would be ridiculous and problematic for me to try to assign races or ethnicities from names and photographs alone. Frankly, it’s a bit awkward to decide whether someone is white or nonwhite from those clues as well, but given that a reader’s implicit biases are probably triggered by them all the same, I decided it was fair enough for this purpose.
Difficult. Difficult, difficult, difficult. Obviously this isn’t something that can even be guessed from a name or a photo, nor is it something that is (usually) included in bios. Thus I could only include authors who have publicly discussed their sexuality or orientation online. Everyone else was technically classified as “unknown”. Several authors had references to a different-gendered spouse in their bios, but I certainly don’t want to assume that everyone in such a relationship is heterosexual. I fully recognize that this is the category which is potentially most incomplete, but again, I can only work with the information available to me.
I made all the protagonist classifications from memory. In cases with three or fewer equally significant protagonists (“equally significant” being basically up to my discretion, but roughly corresponding to whether or not they got a POV at any point), I listed the genders of each. With more than three protagonists and/or ensemble casts, all of which included multiple genders, I classified those books as “mixed”.
I want to highlight a couple of particular cases. Diana Wynne Jones’ Dogsbody features a star, but he spends most of the book in the body of a male dog, so I went with “male”. Naomi Kritzer’s Hugo-award-winning short story “Cat Pictures Please” features an AI as its protagonist, which I classified as “agender”. Ann Leckie’s books Ancillary Sword and Ancillary Mercy present interesting cases. In both books the protagonist is Breq/Justice of Toren, formerly a ship, now confined to a single ancillary body. Now, not only does Breq herself (yes, female pronoun, bear with me) not really have a gender, but the reader is not really aware of the gender of her ancillary body. One of the major conceits of Leckie’s books is that Breq’s language, Radchaai, does not distinguish gender; therefore Leckie uses female pronouns for every character in the book. So…I said “screw it” and classified Breq as “agender”.
I did my best to classify protagonists as white or nonwhite based upon my memory of them; to be entirely honest, however, I might have missed some. I also classified all works which did not explicitly specify the protagonist’s race as “no” (i.e., not nonwhite). It’s all to easy for readers to assume white as a default in those cases, and such a character wouldn’t really count as representation.
Likewise, I only classed protagonists as “yes” in the LGBTQ+ category if the author was explicit about their identity/orientation. That means either stating the words outright or demonstrating clearly, by the character’s words or actions, that they were in some way non-straight. Nancy in Every Heart a Doorway explicitly says the word “asexual” and describes what it means for her (“yes”). Charlie and Aly in The Future Falls reference past sexual relationships with both men and women (“yes”). The Fifth Season has a significant gay character but he’s not the protagonist; she only ever has relationships with men (“no”).
LGBTQ+ identity, in particular, is likely underestimated among authors. All classifications are subject to human error, though I did the best I could with the information I could find. The Hugo Awards likely had some effect on both my average rating statistic and possibly also on the number of male authors in my sample.