Wednesday, March 17, 2010

Statistical Control

Happy St. Paddy's Day! Today is gorgeous. Too bad my stomach hurts...
Now time for class...

http://en.wikipedia.org/wiki/Manipulation_checks


Errors:
Type I  Found result, but shouldn't
Type II Didn't find results, but should have (doesn't get published)

Sloppy >> non-signifigant results
really easy to get n/s results. Random error.


Control is important because higher degrees of control are associated with higher levels of internal validity

Reducing Experimenter Bias
>single-blind
>double-blind


Placebo effect
Rosenthal effect

No comments:

Post a Comment