Confusing/offensive use of "gender" in `starwars` dataset

[dhicks]

The starwars sample dataset includes a gender variable, which currently takes 4+1 values: "female," "male," "hermaphrodite," and "none" (plus NA). There are two problems with this variable and the way that it is coded.

The first problem is that the codes refer to characters' biological sex, not their gender. (For background on the sex/gender distinction, see here: https://en.wikipedia.org/wiki/Sex_and_gender_distinction.) For example, Jabba the Hutt is consistently referred to using masculine pronouns (he/him/his; see https://en.wikipedia.org/wiki/Jabba_the_Hutt). But Hutts are biologically hermaphroditic, with both male and female sex organs (https://starwars.fandom.com/wiki/Sexes/Legends; https://starwars.fandom.com/wiki/Hutt/Legends). So Jabba has a masculine gender, but he's not biologically male.

Fixing this problem is simply a matter of renaming the variable sex; though of course this breaks many examples using the dataset.

The second problem is that, outside of specifically biological contexts, "hermaphrodite" is an offensive term for intersex people. Specifically, the term has been used historically to pathologize and medicalize intersex people (see Anne Fausto-Sterling's Sexing the Body or this Daily Beast piece: https://www.thedailybeast.com/dont-call-them-hermaphrodites). It also continues to be used to objectify intersex people as sexual curiosities. For example, many of the top results for a web search of the term are for porn videos; examples can be seen at https://duckduckgo.com/?q=hermaphrodite&atb=v17&ia=web, though obviously the results at that link are NSFW.

While the term is appropriate in the context of describing the sex of Starwars characters, it's easy for this context to be lost when the dataset is used for examples. The result can be problematic or even outright offensive examples. (Here's the Twitter thread that prompted me to write up this issue: https://twitter.com/danieljhicks/status/1144656643906404352. I also want to note that Mara handled this really well!)

I suggest this second problem can be addressed by switching to adjectival forms. "Hermaphroditic" carries more technical connotations than the noun form of the term; a web search turns up mostly dictionaries, specifically biological discussions, and discussions of intersex people, without any porn: https://duckduckgo.com/?q=hermaphroditic&atb=v17&ia=web. The "none" value, which could be confused with NA, could also be replaced with an "asexual" value, which is more descriptive. (Though potentially confusing in its own way, because that term is also used for a sexual orientation.)

[batpigandme]

Hey @dhicks ,

Thanks for writing this up and for the suggestion that would keep the variable names consistent - ideal! It'd be a pretty brittle test if one exists that relies on exact values in this dataset, but it wouldn't be that hard to sweep for those (I don't think).

@hadley, what do you think about adding this to dev-day issues?

[dhicks]

Just remembered that I wrote up a PR for this and #4455 but forgot to actually submit it

[romainfrancois]

That could be included as part of #3983, probably also related to phalt/swapi#126.

[hadley]

I am a little worried about breaking existing examples. Could we rename the existing gender column to sex (along with @dhicks' suggested terms) and a new gender column that captures the actual gender of the character? (I'm not sure how much research that would involve, and how much is knowable; but it's probably ok even if the column is mostly NA)

[MeganBeckett]

Hi! I just picked up this issue at the Tidy Dev Day :)

To confirm, would one need to determine the actual gender of each character by researching them or use, by default, what is currently specified in gender?

Except for some examples, ie. Jabba, who's sex is hermaphroditic, but gender is probably male? As referenced on Wikipedia: https://en.wikipedia.org/wiki/Jabba_the_Hutt

[hadley]

Yeah, you'd need to determine it by doing some research. I suspect you'll only need to research the gender of characters that don't have male or female sex, but I might be wrong.

[dhicks]

“Feminine” and “masculine” are better labels for gender than “female” and “male.” Jabba is the only hermaphroditic character in the dataset (that was the example I used above), and he’s gendered masculine. IIRC there was one asexual character, and everyone else was female, male, or NA.

[MeganBeckett]

Great, thanks! Yes, I looked up Jabba and saw that he is masculine gender.

For each of the Droids, who's "biological sex" is none, I also looked up their gender as defined in how they were programmed, and have added these in.

[MeganBeckett]

For now though, I've left the current male and female genders as is, rather than renaming to masculine and feminine, respectively. Otherwise I'm not sure if this is too great a change at once that would break examples, etc?

[hadley]

@dhicks oops, yes, I was blanking on the correct terms.

@MeganBeckett I think switching to masculine/feminine would be fine