On 12 May 2004 06:37:30 -0700, [EMAIL PROTECTED] (Paul R Swank)
wrote:

> And doing a Pearson Coorelation and a t-test doesn't tell you the overall
> impact of the error.

If the t-test is *not*  relevant, which may be true for
test-retest, the Pearson can be a more proper measure of 
impact than the ICC  which slightly decreases the reported score.
 - there are extra issues for your study if the variances are
should not be pooled, for any choice of coefficient.

If the t-test *is*  relevant, it can be a warning of a grievous
impact, all by itself;  and that warning is generally masked
by reporting an ICC which may be only slightly less than the 
Pearson r. 

Those are two reasons why the two tests together are better for
*examining*  your data, than looking at ICCs.

Yes, it is the overall impact, and that can be useful for the
*final* statement, especially when a very precise statement of 
overall impact is warranted -- because, for instance, power analyses
are being based on the exact value of the exact form of ICC
that is needed: Same versus different raters; single versus 
multiple scorers.  

And I think it is an over-generalization to prefer an ICC when
the issue is the cruder one of apparent adequacy.  The ICC is
less informative (about means) and less transparent (multiple
versions available to select, all of them burying the means).

[snip, rest]

-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to