Tumblr Science FAQ: The "Null Hypothesis"
I intentionally haven't talked about what hypothesis I'm testing, because that would bias the submissions I get. That's also why I asked for so many.
But plenty of people have talked about my hypotheses, both complaining that I did not post them and speculating about what they might be. And between that and thinking it over, I have decided that I can talk about the "null hypothesis" without introducing bias. So here we go.
What is a "Null Hypothesis", anyways?
The null hypothesis is what you would expect to be the case if you don't find what you're looking for. If you're doing a statistics class assignment to find out if there is a relationship between a person's height and their shoe size, your null hypothesis is that there is no relationship (or at least not a meaningful relationship). If you're doing a study on whether Tumblr moderation and mature-content labeling is biased against trans women, your null hypothesis is that Tumblr moderation is not measurably biased against trans women.
Okay, so what's your null hypothesis, then?
So, we all probably think we have a good understanding of how the Mature Content Label system is supposed to work here on tumblr, right?
By default, post are "for everyone".
The author can mark a post as "Mature Content". This adds a label that displays a clickthrough (for lots of people) that says "the author has indicated...".
Posts can also be marked as "Mature Content" by Tumblr. When this happens, the clickthrough says "Potentially Mature Content", but doesn't say that the author indicated it.
That's my null hypothesis: that the Mature Content Label works the way it says it does.
If it doesn't work that way, well, then I have to talk about what way I thought it would work instead. I need to talk about how I was looking for whether or not it worked that way and what I actually found regardless of whether it works that way, though.