Thursday, April 15, 2004

Otters: Not Actually Fascist

I find this post of mine is being badly misunderstood in the comments section over at Electrolite.
Americans of all ethnic groups are protected equally by the U.S. Constitution, and for good reason. Applying this logic elsewhere, we should have had crackdowns on the civil rights of members of lots of groups. [...] Back to the crackdowns, we might think about cracking down on white males: over 90% of serial killers have been white males. [...]

This means that "these repeated terrorist plots don’t reflect on the 'vast majority of Certain Ethnic Group members, who are peace-loving, law-abiding, loyal citizens,'" and that "evidence that these so-called law-abiding Ethnic Group members covertly support the terrorists among them,"--when based on assuptions of illusory correlation--is hogwash.
Well, um, yes. In fact.

I suppose this is what I get for being ironic among people who don't necessarily know me, or for not festooning my ironic posts with smileys or WARNING: IRONY AHEAD slogans. You all knew better than to take me literally, right? Right?

If not, for future reference: if you find yourself staring at one of my posts and wondering if my brain was somehow taken over by ultra-right-wing mind control rays, please consider the possibility that my words weren't meant to be taken literally.

Incidentally, how hopelessly geeky am I? My first response to this paragraph:
If you added up all the members of your "certain ethnic group" who have been arrested (or are wanted for) for acts of terrorism, then compared that number to the total population of members of the group living in the country or the world, and then ran a very basic test of statistical significance called a t-test, the test shows that a relationship between terrorists and your "certain ethnic group" has NO statistical significance.
was to start picking apart the guy's understanding of statistics. In fact, if all you know is the population of a group and the number of terrorists who are group members, you don't have any means of testing whether there is a statistically significant relationship between group membership and terrorism. You would need to be able to compare the ratio of terrorists to group members with the ratio for other groups. And even if you did have rates for all ethnic groups, you would need to use a chi-square analysis, not a t-test, because the comparison as described violates the statistical assumptions underlying the t-test, such as -

Okay, never mind. Sometimes I'm so geeky I can't stand myself.