On 8 Dec 1999 15:53:34 -0800, [EMAIL PROTECTED] (Magill, Brett)
wrote:
> Mike,
>
> With randomization pre, it is not necessary to take a pre-intervention
> measurement. Test the difference in confidence following the training. If
> it is significant, there is a difference. Decide what direction it is in
> and attribute the difference to the training. You can make this attribution
> because of random assignment even without pre-measure.
< snip, citation >
To be more complete: the one group was
a) assigned;
b) trained, we presume; to varying levels of skill;
c) rated -- not at all 'blindly' in respect to the training.
If there was a huge sample and not a difference, that would be curious
in one way.
If there was a difference that was totally uncorrelated with skill, or
uncorrelated with the amount of skill learned (two different things,
perhaps), that would be curious in another way, and could suggest a
superficial effectiveness. ( - something like a Hawthorne effect, the
possibility of which, people should be aware of.)
--
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
