Monday, August 03, 2009

AYP NM: Making Sense of the Numbers

Having had a chance to look a bit deeper into the test scores released today, I think it's important to first mention a few positives before assailing you, dear reader, with the nunchucks of statistics that follow.

Speaking specifically of APS schools, I see two elementaries that stand out (btw, I have no connection to either of them, and don't even know anybody who works there).

Alvarado Elementary is a North Valley school, pretty darn close to Sadie's Restaurant. Alvarado was remarkable this year because over 50% of its "Students with Disabilities" SWD (i.e., Special Education kids with half-time SpEd service or more) were proficient in Math. Their figure of 54% is higher than many school's non-SWD populations. It's really high.

Combined with eeking out the SWD score in Reading, Alvarado Elementary was a very rare incidence of an APS school that both "passed" AND had enough SWD kids to count toward "passing". More about exactly how rare an occurrence that is below.

La Mesa Elementary is in what can be considered a "tough" part of Albuquerque, just East of the NM Fairgrounds. Its population of "English Language Learners" (ELL) accounts for over three-quarters of the entire tested student body. These ELL kids were proficient in Math at a 51% clip. Now it's true La Mesa did "pass" overall this year solely because of "safe harbor" provisions, but that Math number for ELL kids is, again, remarkable.

There are other successes in the dense thicket of scores released today, but these two stand out to me.

Now on to the nunchucks.

In case you don't already know, the New Mexico Public Education Department doesn't consider a testing subgroup as statistically relevant unless there are 25 members of a particular subgroup tested at a school. For example, if School X tests 24 "Students With Disabilities" that subgroup doesn't count. If School X tests 25 SWD kids..the subgroup counts.
  • Number of schools within the APS umbrella reporting scores today: 168
  • Number of "passing" schools: 39
  • Number of "passing" schools with 25 or more SWD students tested: 4 (Alvarado, Monte Vista, S.Y. Jackson, Zia...all elementary schools)
  • Number of these four "passing" schools that also had 25 or more ELL students tested: Zero
In other words, zero APS schools "passed" who had the statistically relevant number of test participants in the areas of both "Students With Disabilities" and "English Language Learners". Zero.

Keeping in mind that a school had to have 25 Full Academic Year students (i.e., students who attended the same school for most all of the testing school year) participate for a subgroup (e.g., Students with Disabilities, Hispanics, etc.), the following factoid is also slightly interesting:
  • Number of schools within the APS umbrella reporting scores today: 168
  • Number of "passing" schools: 39
  • Number of "passing" schools with between 17-24 SWD students tested: 14
  • Number of "passing" schools with between 17-24 SWD students tested who were on a pace to "fail" in the area of SWD: 10
I'll leave the breakdown in regards to ELL students to others.

The above illustrates a fact and raises a paranoid concern. First, it's easily apparent the number of schools "passing" is bogus. So much is tied to whether a school has 25 test participants in a subgroup that the idea of "passing" is a joke.

Second, I would NEVER suggest that some elementary schools are doctoring their number of SWD kids in an attempt to stay just below the 25 student threshold. I would NEVER say that...but if you look school by school, and keep in mind the exponentially more convoluted screening procedures for Special Education these's not terribly hard to turn the above factoids into paranoid speculation.

One only wonders how much more rampant such paranoid speculation would be if test scores were ever truly tied to something like "Merit Pay".

Okay, enough for today. Tomorrow I think we'll again tackle what to do now, from tearing down the test itself and rebuilding a fairer system, to responding to critics who seem to think this testing means one thing when, in fact, it very much means something else.


Anonymous said...

La Mesa teachers rock!!

Treyomo said...

School districts would never doctor statistics...just like they would never make a single day effort to chase truant kids into school on the day census is taken that determines school funding, then ignore the problem the other 179 days.

Anonymous said...

It is federal, No Child Left Behind mandate that subgroups are reported only if their numbers are 25 or greater. It is not just a PED mandate. Get the facts straight.

jscotkey said...

Anonymous (the last one): Au contraire on the number of 25 or greater thing. States determine the numbers for subgroups to count or not count.

There's a great little study about this that I've been raving about for some time now:

My favorite example is Wisconsin, which requires a "student with disabilities" subgroup size of 50. Some states are lower, others higher, with a few having different ranges for one subgroup versus another. For instance, Ohio requires 45 for "Students with Disabilities" but 30 for other subgroups.

There is a great deal of play in the numbers, actually, as well as in the "confidence intervals". I hope you stick around "Anonymous" and see this study, and can take a gander at it yourself.