Sunday, August 06, 2006

Intro Stats meets Dykes to Watch Out For

or, why I shouldn’t take night classes
with apologies to Alison Bechdel

(These are actual notes I took in an actual night class about survey research methods. I don't know how closely they relate to what the prof was actually saying, but I did get an A so they can't be that far off. If you don't know much about statistics, and none of this makes any sense at all, you can look it up in Wikipedia. And if you think my examples are out there, you should see what the Wikipedia folks came up with.)

null hypothesis: we are not in a relationship.

alternative hypothesis: we are not not in a relationship; there may in fact be a relationship going on.

data you can use to reject the null hypothesis: her panties in my underwear drawer. his skim milk in my fridge, and my whole yoghurt in his. the u-haul brochures that keep showing up in our mailboxes. our white shirts dyed pink by my red hanky.

type one error: erroniously rejecting the null hypothesis. this type of error is particularly embarassing. "Dude, I can't believe I said that. I totally thought they were in a relationship."

type two error: erroneously failing to reject the null hypothesis. this type of error is less embarassing, though no one's quite sure why. "Dude, I am certain they are not in a relationship. There is no evidence of a relationship. Surely it's just random chance that they keep disappearing together."

power (statistical): how savvy is your relationship-dar? and how hot are you when you're the dyke that always knows first?

significance tests: "They are totally a couple, p<.001." or, "They may be in a relationship, but it's not a significant one, and if we were to retest in six months we might find no relationship at all."

No comments: