In article <[EMAIL PROTECTED]>,
Alan Zaslavsky <[EMAIL PROTECTED]> wrote:
>The following article may be of interest to some of you who are trying to
>get across the notion of reliability, particularly those who are teaching
>H.S. or young college students who have recently gone through high-stakes
>achievement/competency testing programs.
>[excerpt:]
>
>Are standardized-test scores, on which many schools rely heavily to
>make promotion or graduation decisions, more indicative of true ability
>than a ballplayer's weekly average?
>
>Not really. David Rogosa, a professor of educational statistics at
>Stanford University, has calculated the "accuracy" of tests used in
>California to abolish social promotion. (New York uses similar tests.)
A fundamental flaw with this article is that is assumes that school
promotion decisions are some sort of contest, in which winning is
good, and in which every student ought to get a fair chance at the
prize. This would be the case if we were talking, say, about students
competing to get into a school with better teachers. But I think
that's not the context.
We are instead talking about deciding what educational program is best
for the student in question. It is NOT the case that promoting the
student is necessarily the best thing for the student. Schools are
just as derilict in their duties if they promote students who will be
lost in the higher grade as if they keep back students who would do
better in the higher grade.
Furthermore, for students on the borderline, it (by definition) does
not make much difference whether they are promoted or not. For these
students, the advantages of being exposed to the more advanced
material are just about balanced by the disadvantages of not having a
firm grasp of the prerequisite material. So it makes very little
difference if a student whose true ability is slightly above the
cut-off scores slightly below the cut-off on the test. (Assuming, of
course, that the cut-off is set at the right level.)
Radford Neal
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
