Somewhat timely as I discuss this very topic with Tobias Buckell in the next episode of Novy Mirror podcast, Vylar Kaftan posted this analysis of diversity in her own fiction. I decided to do something similar, just to see how I was doing. (Yes, I know it's just a way to procrastinate.)
I should point out that my statistics will be biased toward my own demographic because I often use myself as the main character--sometimes in disguise, sometimes not. If I were to remove those stories, I think the distribution might change. Also, recurring characters were counted as many times as the number of stories where they appear. This also biases the results some.
I counted characters that play a major role in the story, not just the POV character. I came up with a count of 154 characters across all my short stories and the completed novel. Work in progress is not counted.
Gender - heavily biased by my tendency to use myself as a point of view character.
Male = 67%
Female = 27%
Other = 6%
Race - Again, biased by using myself as a character.
Caucasian = 48%
Native American 0.5%
Using the assumption that characters default to white
White or implied white 66%
Person of color 17%
Unclassifiable 20% (= more than 100% due to rounding errors)
Unknown = 60%
By similar logic, characters default to straight. 100% No rounding errors here.
Ruling 0.5% (FDR if you must know)
Conclusions: The stories where I use myself as the POV character really skew the results toward my own demographic. I think I use race effectively, though I could improve. Gender tilts toward male for the same reason, but I did notice a tendency for my stories to use two males and one female. Often there is a male-female protagonist team against a male antagonist. However, in my novel the primary antagonist is female. Age looks to be almost a normal distribution, so not a bad use of various ages. I tend toward middle-class characters, skewed toward the lower middle class more than the upper middle class. Or, there is no indication of social standing at all. For disabilities, I think some of those I counted as "able" might qualify as disabled under Vylar's rules. In terms of sexual orientation, well, what can I say? I need to be more aware of this possibility in order to create realistic story people. However, I think that several characters in the "unknown" category could easily be gay. It's just that sexual orientation was irrelevant to those stories.