At 01:49 AM 1/17/01 +0000, Elliot Cramer wrote:
>There seems to be some confusion about what regression to the mean
>means. Noone is penalized (or advantaged) because of regression to the
>mean. You ALWAYS have RTM in a population whether everyone improves or
>gets worse. It is a property of standardized scores only for a
>population. The simplest explanation is in terms of the regression
>equation for standardized scores
well, let's just think about this ... let's say that i give a test ... test
1 ... and there is reasonable spread ... and it covers material in units 1
to 5 ...
i happen to get lucky and get near the top ... AND THE INSTRUCTOR IS GOING
TO BASE GRADES ON PERCENTILE RANKS ON THE TESTS
well, after we do units 6 to 10 ... we get test 2 ... with reasonable
spread ...
so, what kind of a correlation would you expect between the two measures?
could be .7 perhaps ...
now, generally speaking ... while my score might still be near the top ...
RTM suggests that ON AVERAGE ... the people with the highest PERCENTILE
RANKS will not have as high an average percentile rank on test 2 ... so,
for me ... even though my RAW SCORE MIGHT BE HIGHER THAN ON THE FIRST TEST
... if the instructor is going to base my test grade on my percentile rank
... i could easily get a lower grade on test 2 than test 1 ...
somehow, i think this is penalizing me for the fact that there is not and
cannot be a perfect correlation between the two measures ...
what happens to the ones at the bottom ... relative to what happens to me??
as i have said before ... regression to the mean is a feature of relative
position ... and not necessarily RAW scores ...
THAT is what we need to keep straight when thinking about the impacts of RTM
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
