Careful, Tim,
this is where the State of Massachusetts came in. If you select a group
based on their position at the bottom 10% or top 10% on a test, and then
retest, _even the next day_, we can expect the bottom 10% to improve!
Fantastic! That was one heck of a day's education! IN addition, the top
10% will drop. guess we'd better get on those teachers' cases, for
doing such a lousy job. Especially all in one day!
This is Galton's 'regression toward the mean,' and it will bite you at
least every time you select the study groups based on their ranking on
the test, and then retest the groups.
Without going into the details (see refernces given before), I suggest
you select groups based on demographics, and other non-test graded
variables.
Ask exactly what you intend to do with the results - assess students,
teachers, or schools. It makes a big difference how you adminster, and
what you have on, the tests. If the test cannot answer the technical
quesiton you ask, then there is no sense running the numbers for analysis.
Jay
Tim Victor wrote:
> Here's the scenario. We have a years worth data on several school
> districts consisting of:
>
> ° scores obtained from the beginning of the school year
> ° scores obtained from the end of the school year
> ° ordinal index of beginning of the school ability, e.g., at risk,
> mainstream, gifted
> ° grade
> ° demographic variables
>
> A student's score would determine group membership.
>
> We would like to use these data to predict the proportion of students
> in each group who will move from the current group (k) to the k+ith
> group.
>
> I'm thinking Poisson regression might be the way to go here. I'd like
> to hear/read others' thoughts.
>
> Thanks in advance.
>
> Tim
>
>
> Sent via Deja.com
> http://www.deja.com/
>
>
> =================================================================
> Instructions for joining and leaving this list and remarks about
> the problem of INAPPROPRIATE MESSAGES are available at
> http://jse.stat.ncsu.edu/
> =================================================================
>
>
>
--
Jay Warner
Principal Scientist
Warner Consulting, Inc.
4444 North Green Bay Road
Racine, WI 53404-1216
USA
Ph: (262) 634-9100
FAX: (262) 681-1133
email: [EMAIL PROTECTED]
web: http://www.a2q.com
The A2Q Method (tm) -- What do you want to improve today?
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
