Whoops! We CAN do better!

In November, I wrote a post about how the state's goal for school's API scores was 800, and how it was impossible for more than 40% of the schools to meet that goal. You know what? I was wrong... sort of. There is a way for 60% of school's to reach that target, but I still maintain that it's gonna be nigh impossible to get there.

Why? Well, imagine that there are 100 students in the country, and we rank them according to their test scores into quintiles (what's a quintile, you ask? Check this post out for an explanation). That would mean that 20 kids got a score of 200, 20 got a score of 400, 20 got a score of 600, 20 got a score of 800, and 20 got a score of 1000. You can see that only 40% of the students got 800 or above. The school's API, however, depends on the average of all of the students in the school. So it's possible to arrange the students so that the average is at or above 800 for more than 40% of the students.

How? Well, it's those kids that scored 1000. We can use their extra points to balance out some other kids who scored under 800. The simplest example would be to match up one student with a score of 1000 with one student with a score of 600. If those two students made up the entire school, that school would have an average API of 800. See? Easy, huh?

Unfortunately, it's only 20 percent of the students that have these extra points, so if we make 40 % of our schools populated with exactly half of their students in the 600 range and half in the 1000 range, we can bring up those schools to 800 as an average score. Keep all of the students who scored 800 together and their schools keep their 800 score, giving us another 20% of our schools meeting the target. Those kids who scored 200 or 400? Well, the problem is that we'd have to waste 2 or 3 1000 scoring kids on each one of those students to bring the average up to 800, which would mean less schools overall would have the average that we want. Sorry, bottom-percentilers. You lose, but America wins... right?

Not really. Notice that the only thing I'm doing is rearranging which schools these kids are going to. No increase in learning or improvement in instruction has to happen for more schools to come up to the statewide goal, we just need to mix the students up a little bit.

Actually, that's not a bad idea. I am not alone among educators who think that a heterogeneous population in a school and in individual classrooms makes for a better learning environment and improves learning for all students, even those at the top (of course, this won't necessarily improve test scores, since they don't measure learning, just ranking). So why do I say it's still impossible? Well, you'd probably be able to convice parents and students at a school averaging 600 that it's a good idea for some students from their school to be transferred to the school across town that averages 1000. The problem comes when you try to get parents and students (mostly parents) of the students at school 1000 to move over to school 600.

Well, I have a proposal that could actually make the standardized testing system marginally worthwhile. Not worthwhile enough to keep doing it the way we do, but less of a complete waste of time, energy, resources, and money. What if a federal mandate demands that all schools must be made up of a student population that has an average API of exactly 600? Testing would be given in 3rd grade, 6th grade, 9th grade, and 12th grade. Transfers would only be permissible in the 4th grade, 7th grade and 10th grade, and those transfers would have to rebalance the API averages in those schools back to 600. (It's cruel, ineffective, and nonsensical to administer standardized tests to kids before 3rd grade. They can't read well enough or sit still and focus long enough to make it worth the effort.) That way, every "Elementary B" school ("Elementary A" being Kindergarten through 3rd grade) would start off with the same average level of test-takers (notice I don't say that they're at the same level as far as actual knowledge or skill... just test-taking), and then the test that they take 3 years later would actually show whether their test-taking improved as a result of 3 years at that school or not (again, this wouldn't necessarily tell us anything about their learning during those 3 years, just their test-taking ability). Likewise for 7th -9th grade "Middle Schools" and 10th through 12th grade "High Schools." Any schools that scored above 600 would have improved their students test taking ability relative to the average improvements across the nation. Lower than 600? Well, that would be bad, wouldn't it?

It still wouldn't tell us much about whether the kids are learning anything, but at least it would be a fair comparison of the schools' test-prep abilities. It's a far cry from assessing actual knowledge, but isn't it better than testing for socio-economic status and race, which is what the tests as they're currently set-up do test for?

- "That's impossible, no one can give more than one hundred percent, by definition that is the most anyone can give."