f:\doc\web\98\07\normbase.txt


Mail it to 


Arthur Hu
12422 107th PL NE
Kirkland WA 98034




Problem with norm based tests is that it frustrates the hell when
activists are told that on average, most people are average or
better.  Can't show failure with that. That's why you have to toss
the bell curve. You need a game where all students can fail, not just
the poor black kids.


Percentiles are fine when applied to individuals and schools - if a
school is at 10, you know it's pretty horrible. If a school is at 90,
it's pretty damned good. If a state is 50 or a little above, it's
nothing to get outraged about, since states are pretty much
approximations of the national average anyways. Most cities are at
around 50, which you might expect since a so much the population
lives in cities, and it doesn't tell you much because it's usually a
mixture of elite whites who raise the average, and minorities that
drag it back down.
 
State rankings are pretty much useless when the district or
particular school you go to has much more to do with any individual's
performance.


You can see the question of setting one standard of "How good is good
enough" is a scam if you think about setting what "percentile" point
of a norm-referenced test is acceptable performance. There is no line
above which performance is acceptable and below which the person is
completely worthless. How good depends on "for what"?  MIT takes the
top 1%. UC Berkeley takes the top 5%. UC system takes top 15%.
College goers are generally the top 66%. HIgh school grads are the
top 85%.  Every selective institution sets its own "bar". There is
and cannot be one bar for all.


The old way is to pass anyone who achieved any amount of learning and
played the game by the rules, and give anyone who did better
corresponding grades. Now it's a pass / fail game, but they're
failing all but the top 10-20% because somebody thought that if
anybody can be smart enough to get into M.I.T, then everybody must be
able to.


When states like Washington assemble committees of volunteers who are
"facilitated" to a "consensus", the result is a "standard" that is
worthless for all purposes when it is deliberately set to have no
relationship to actual performance levels currently observed in the
population.


It's the notion of one high standard for all that's bogus, not states
or districts posting above average scores.


From:           	Cmr1234@aol.com
Date sent:      	Wed, 30 Sep 1998 23:50:33 EDT
To:             	arthurhu@halcyon.com
Subject:        	Re: Achievement Goes Beyond Race and Poverty


> Arthur:  Are you aware of the work of Dr. John Cannell, who found that all the
> standardized achievement tests (Iowa, SRA, etc.) have been corrupted so that
> every state gets norms that show its students to be above the national
> average!!??  
> 
> He wrote a book that came out about 9 years ago.  It was reviewed favorably by
> Chester Finn and others, but Congress never did anything about it.  I have an
> articl I can send you if you give me your mail or fax info.  
> 
> Meanwhile, keep holding their feet to the fire!  
> 
> Charlie Richardson