| *needs a fandom, please icon* |
[16 Nov 2008|11:29pm] |
Okay, it's times like these that I'm glad I'm not real active in my fandoms. I mean, I like to pretend that I'm a prominent author, but I don't really do meta and I avoid discussion forums like the plague, because I don't like indulging in thinky-fandom-talk with people I don't personally know because it leads to stuff like this. So. Fandom please!
To take my mind off your shenanigans I am going to perform a significance test on my last iTunes meme to prove that iTunes is either psychic or psycho and that I am indeed learning something in my Prob/Stat class.
So. This is an observational study. The explanatory variable is iTunes being psychic; the response variable is whether the song that comes up in response to a question is appropriate or not; both are categorical binary. Technical conditions are met. It's a simple random sample, and .45 x 31 is greater than ten, so the sample size is large enough.
Right! So, my data states that the statistical likelihood of getting a song appropriate for the question is .45. Logic says that the statistical likelihood of getting a song appropriate for the question is .5. Therefore, I shall perform a significance test to see if getting data as extreme as the data I got is statistically likely if the true population mean is .5.
Null hypothesis: the true mean is .5.
Alternative hypothesis: the true mean differs from .5 (ie, iTunes is pyschic or psycho).
Test statistic: -6.936
p-value: <0.0001
significance level: .05
The level of significance is high enough that the true mean is highly unlikely to be .5. I reject the null hypothesis at the .05 level and accept the alternative hypothesis, and given that the test statistic is negative, I would say that iTunes is not psychic but psychotic.
See? I learned something! Applaud me even if that was gibberish.
|
|