On Tue, 3 Oct 2000, dennis roberts wrote:
> michael ... the model of simple regression to help define
> over/underachievement is not mine ... it is just the standard way it is
> defined ... particularly within psychology ... there may be other
> connotations but, i believe i am correct in saying that this is the
> commonplace technical one ... i believe that robert thorndike was
> responsible in making this distinction in reference to gaps between
> predicted achievement and actual achievement ...
OK.
What I find interesting is that this gap between predicted and actual
outcome is much more widely used. The diagnostic tests for learning
disabilities (as I understand them) and legal definitions (again as I
understand them, which could be wrong) of bias or discrimination are based
on poor fits of some model to actual outcomes.
If I understand your question, and I may not, you are wondering about the
legitimacy of such practices. And, if my understanding of learning
disabilities, bias, and discrimination are correct, this is an important
question.
>
> if we did not use predicted achievement ... then low ability folks would
> always be overachievers and high ability folks would be underachievers ...
> just because of the regression to the mean phenomenon ...
>
> as for bringing in motivation ... while i know that this variable impacts
> on effort and learning ... i am not sure i want to include it in the model
> for over and underachievement (though i am not sure about this) since ...
> one might be able to do something because of one's ability ... but, clearly
> not motivated ... so, we get a depressed expectation (if motivation is in
> the model) of achievement NOT due to one's capacity (whatever that means)
> to perform the task but rather, ones "stubbornness" (if i might phrase it
> that way) to actually perform the task ...
>
>
> however, since i am not fond of this concept anyway ... i am not sure i
> want to pursue a more accurate model for predicting achievement IN the
> context of comparing it to measured ability ...
>
> i am certainly interested in a better model for predicting achievement ...
> just not over and underachievement ...
Then what is the problem with including motivation again? It should
improve your match between observed and predicted achievement. Or is it
that you are looking for a better way to predict over/underachievment a
priori rather than waiting for some period of time to see where we've made
prediction errors?
Perhaps we could use some sort of attitudinal measures to determine who is
going to be an over/underachiever absent any knowledge of ability. (I
suppose we'd have to establish the validity of this approach first, but it
seems likely we could identify motivated and nonmotivated individuals and
predict over/underachievement based on that.)
>
> i do know that schools and school personnel ... make a big issue out of
> this ... particularly underachievment and, i think in general, this is a
> dangerous and not really a validated notion ...
Which notion specifically? The notion of using discrepency to define
underachievment, or underachievement as a notion? The following paragraph
suggests the former rather than the latter, but I want to be sure.
>
> we are using these GAP measures (between predicted and actual achievement)
> to define over and underachievment and, we know that difference values are
> notoriously unreliable ...
Lack of reliability doesn't make the concept invalid, just noisy. Could
we use better measures of ability and achievement? I'm guessing yes. But
isn't over/underachievment as a concept primarily a discrepancy?
>
> and as i said before ... this definition is based on the ERRORS in the
> model ... not some "latent" characteristic of the student ...
That seems right. We define over/underachievers as folks who are
discrepent relative to the "standard model". We identify the outliers and
label them for special attention. If I understand you consider the
existance of these outliers as evidence of a problem with the model,
whereas the standard practice is to consider them as unusual individuals.
I'm still not sure where my thoughts are headed, but I'm definitly not
seeing the same big problem as Dennis on this one.
Mike
>
>
Michael M. Granaas
Associate Professor [EMAIL PROTECTED]
Department of Psychology
University of South Dakota Phone: (605) 677-5295
Vermillion, SD 57069 FAX: (605) 677-6604
*******************************************************************
All views expressed are those of the author and do not necessarily
reflect those of the University of South Dakota, or the South
Dakota Board of Regents.
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
