Heteroscedasticity of correlates

It isn't the quality of what you say that matters so much as how many people look at it - Blogopotamian credo.

In the interest of adding an increment to the excrement found in Blogopotamia I have decided to dedicate my next series of posts to obscure keywords and relate them to current events. Today's keyword is "heteroscedasticity," which literally means varying scatter widths (no not scat widths). It's a word derived from the Geek words for things that don't relate to each other the same way all the time.

It's really a statistical term used to describe scatter plots like the ones above.

Scatter plots show the actual (as opposed to the perceived or imagined) relationship between two measurements such as poverty rates and test scores (actual) or student to teacher ratios and test scores (imagined).
If the scatter width (generally measured on the Y axis) varies it tells you there's something else going on in the relationship at higher or lower values of X.

So you can see that at higher poverty rates there is much more test score variation than at lower poverty rates. The relationship, in other words is heteroscedastic, or as the British would say "heteroskedastic" (which is pronounced the same, but said it a funny voice that's somehow both pompous and understated because they say things like that all the time.)

The relationship between Student to teacher ratios and performance on the other hand is not heteroscedastic, but in fact homoscedastic. (It's not that strong either and it tends to be negative despite what you'll hear from people who think it's obvious that kids learn more in small classes.)

Heteroscedasticity is like a plate of shrimp(Repo Man 1984). Once you hear about it you will see it all around you like when you are eating your Cheerios and say, "Hey that's heteroscedastic" to which your wife says, "Are you speaking Geek again?" To which you say, "I need to mictorate" to which she says, "Be sure to elevate the U-shaped buttocks cradle!"

But then you go outside instead, because you like to pee in the snow and look at the patterns. After you finish trying to spell out "Scooter Libby was framed" in a snow bank you look at the cursive trail in a new light. You see that your stream was much steadier at the beginning than at the end when it was fairly wild and scattered all up and down the snow bank. You look at your piece of work and say knowingly to yourself: "Now that's Heteroscedasticity!"