Working Group for Nigem Methods Class: What is a Regression Artifact?

Tuesday, April 05, 2005

what is a regression artifact?

anyone?.... anyone?.... anyone?.....

Clstal said...: I got an answwer for you! Yay google!

http://www.naropa.edu/faculty/johndavis/prm2/exper3.html

It's those oddballs who are freakishly good or bad at whatever you're measuring throwing off your damn results.; 10:08 PM
harvestorm said...: Gotcha! That's a good link, BTW. I guess I was just thrown by the word "artifact" (it reminds me of something churchy); 10:13 PM
Clstal said...: James,

Reading what you wrote and rereading the link, I think I oversimplified in my example.

It'd be having people who are really good or really bad in your sample and the drift that would happen because there is simply no other direction for them to go. They would throw off your results, but not in the way I said.

If they were terrible - in retesting, or over time, they'd get better, simply because they can't get worse. If they were STELLAR they would have to get worse over time. It's the difference that'd throw you off. (I'm thinking delta v for change in volume and mixing my chemistry in here).

If you only measured em once, you wouldn't get this problem.; 10:28 PM