\doc\web\98\06\wasl710.txt


"The WASL is a test of what 4th graders are expected to learn and
know by grades 4". (advertised)


"The WASL is a test of what 4th graders will expected to learn and
know by grades 7 and 10". (the truth)


9/21/98
Arthur Hu


Olympia - WASL Intentionally Puts G7 and G10 questions into G4 test?


I went down to Olympia to bug Gordon Ensign and his buddies on why
they've put G7 and G10 questions on a G4 test, Washington Assessment
of Student Learning. He and other OSPI people are evidently now
saying I might be right, it's in the G7 and G10 areas ,but we have to
put in questions to challenge kids above the standard. 


That might make sense for a norm-referenced tests, but not a
one-size-fits all OBE "all will suceed at the same level" test.
Nothing in the documentation - the benchmarks or the draft
specification which says how the test will be designed calls for a
mix of G7 and G10 problems. The test is sold as a test of what
"students should be able to do by the 4th grade" Not when the
benchmarks say they should be able to do by G7 or G10. So the truth
is that they are giving 4th graders a test of what they are expected
to know by grades 7 and 10.


Most tests like the NAEP specifically state that content at higher
grade levels like independent probability, ratio and direct
proportion will NOT BE ASSESSED at grade 4. That's the national
standard. Did they make a a conscious decision to go against the
national standard or are they just covering up for the fact that my
assertion that they never checked in the first place is true?


Joe Zarelli also observed that the 3 day "standard setting" process
is a scam because the "standard" does not exist until AFTER the test
is designed, and given to the kids.  Then the scorers have a big
consensus party to decide what constitutes a "correct answer" for
each question for all possible responses and levels 1-4.  They
admitted that scoring rubrics are "vague" because they cannot
anticipate "unexpected but valid" responses. They say "that's the
good thing about performance based testing" when in fact that is the
major flaw about it.


The standard setting "commitee" never got to pick which questions are
on the test, they don't even see the test until after the kids take
it. They say that there are other experts who evaluate the test, but
we can see that they've clearly given the green light to G7 and G10
problems.


Zarelli asked if his child studied and knew everything they were
supposed to know at 4th grade if they could ace the test. I wanted to
raise my hand, but it was closed to public input. The fact is that if
you completely master everything in the 4th grade benchmarks, the
fact that over 40% of sample problems require 7th and 10th grade
skills, and you have to get 65% right to "meet" the standard means
you have to get 100% of 4th grade problems, plus some over grade
problems to pass. No wonder only 20% were able to pass this test. 20%
is the same proportion of kids who go on to the University of
Washington or U of California. The expectation is that all children
will perform at this level. Yeah, right.  You and what data says
that's going to happen?


Once the tests have been scored, THEN tests are given to the
committee who line up the questions from the easiest to hardest,based
on how they were scored who "bookmark" where they think the line
should be. The next year's test uses a "scientific" equivalence
method, which sounds great, but they still have no way to demonstrate
that the reason 50% more kids scored above the standard wasn't simply
because they got a test that was 50% easier.


There is no document that shows exactly what questions are below or
above the standard. Logically speaking a G7 or G10 problem should
"meet the standard" even if the solution is incorrect since they are
not expected to have to answer the question to meet the standard. But
the scoring process (by the way, they consider 10% disagreeement to
be an acceptable level of error between graders, compare that to 0%
with machine scoring) only judges 1 to 4 assuming that aswering the
question correctly "meets" the standard.


There is no way to include a question that can be scored as "meeting
the standard" when the standard is that the skill is not required at
the grade level.


So the WASL remains completely closed to public input, it is only
open to those invited to participate in its consensus creation, a
process designed to produce precisely the pre-destined outcome.
Gordon Ensign and Terry Bergeson both refuse to acknoweldge even the
possibility of errors in the process, let alone errors as large as I
have alleged.