Education's California Adventure
Something strange is afoot in California -- or is it? The L.A. Times reports:
(Note: A value-added growth model, if you hear that term, refers to a growth model that tries to isolate student progress to a specific time and agent, usually with the help of statistics. The API, for reasons that will soon become clear, is not a pure value-added growth model)
California's API, which is based off of but separate from the state tests which are used for federal accountability, is a growth model of sorts. The API essentially synthesizes a school's performance on various tests, weighting each test differently (so that the math and reading California Standards Test are the most important, for example), and comes up with a score. The scores range from 200 to 1000, with 800 the statewide target. Each year, schools have an API growth target which is 5% the difference between their current score and 800. So, a school with a current API of 600 has a growth target of .05 x (800-600), or 10 points. Thus, schools further away from the standard have to make more annual growth.
Now, here's what you need to know about the API: It's not impressive. There are no real consequences to a school for failing to make its API growth target (go go gadget accountability!), and no substantive incentives for succeeding. Moreover, since California doesn't have unique student identification numbers yet, the API isn't even a true growth model -- it looks at one class of students compared to the next year's class of students (e.g. 2004 fourth graders vs. 2005 fourth graders) rather than tracking groups. Worse yet, each relevant subgroup -- racial, etc. -- only has to make 80% of the school's overall growth target to meet API regulations. In other words, a school's Latino students can only make 8 points of growth in our above example so long as the white students compensate for them. Soft bigotry of low expectations, anyone?
Growth models have a lot to be said for them if designed effectively and applied correctly. The API has some good aspects, but is in the aggregate weak. To be sure, status models like AYP also have plenty of flaws, such as ignoring the progress of previously low-achieving schools and holding schools accountable for students whose educations may have been previously slowed by other educative institutions. But let me say this -- Asking that one-quarter of your students be proficient in English language-arts is not unreasonable! Even in such a multiracial and multilingual state as California, 25% isn't exactly reaching for the sky. It's barely reaching for the ground.
So, on this count I'm squarely in the corner of the big bad federal government. That more schools are meeting their API growth targets is good news (though i'd like to see how many schools at the low end made their targets vs. schools already in the 700s), but adulation should be tempered by an understanding of just what the API is. California's defensiveness is like trying to claim victory after plugging one hole in a swiss cheese boat. The untold tens of thousands of California's kids who still can't read are the real story.
Nearly one-third of California public schools won praise today for meeting state achievement goals on test scores even while they were branded as failures for missing a federal gauge of success...What we have here is a case of a growth model going up against a status model. This is a topic near and dear to my heart, as I spent many weeks working on it during my EdTrust internship. Simply put, a growth model looks at a student or group of students' progress over time. A status model just checks to see if student achievement is above or below a certain line.
Campus leaders were left to decipher the differences between California's Academic Performance Index, which rewards incremental test score gains, and the federal No Child Left Behind law, which requires schools to clear a rigid achievement bar that rises regularly...
State education officials said they were pleased that 81% of the schools had met their state improvement targets, up from 64% last year.
But the officials were not happy about the results under the federal system: nearly 2,300 schools that met their state targets still fell short of the No Child Left Behind goal. More than 5,100 California schools met their federal goals.
That's because the federal bar rose for the first time this year, leaving many campuses unable to reach it.
To pass the federal bar this year, elementary schools and middle schools had to raise at least 24% of their students to the proficient level in English-language arts, up from nearly 14% last year.
(Note: A value-added growth model, if you hear that term, refers to a growth model that tries to isolate student progress to a specific time and agent, usually with the help of statistics. The API, for reasons that will soon become clear, is not a pure value-added growth model)
California's API, which is based off of but separate from the state tests which are used for federal accountability, is a growth model of sorts. The API essentially synthesizes a school's performance on various tests, weighting each test differently (so that the math and reading California Standards Test are the most important, for example), and comes up with a score. The scores range from 200 to 1000, with 800 the statewide target. Each year, schools have an API growth target which is 5% the difference between their current score and 800. So, a school with a current API of 600 has a growth target of .05 x (800-600), or 10 points. Thus, schools further away from the standard have to make more annual growth.
Now, here's what you need to know about the API: It's not impressive. There are no real consequences to a school for failing to make its API growth target (go go gadget accountability!), and no substantive incentives for succeeding. Moreover, since California doesn't have unique student identification numbers yet, the API isn't even a true growth model -- it looks at one class of students compared to the next year's class of students (e.g. 2004 fourth graders vs. 2005 fourth graders) rather than tracking groups. Worse yet, each relevant subgroup -- racial, etc. -- only has to make 80% of the school's overall growth target to meet API regulations. In other words, a school's Latino students can only make 8 points of growth in our above example so long as the white students compensate for them. Soft bigotry of low expectations, anyone?
Growth models have a lot to be said for them if designed effectively and applied correctly. The API has some good aspects, but is in the aggregate weak. To be sure, status models like AYP also have plenty of flaws, such as ignoring the progress of previously low-achieving schools and holding schools accountable for students whose educations may have been previously slowed by other educative institutions. But let me say this -- Asking that one-quarter of your students be proficient in English language-arts is not unreasonable! Even in such a multiracial and multilingual state as California, 25% isn't exactly reaching for the sky. It's barely reaching for the ground.
So, on this count I'm squarely in the corner of the big bad federal government. That more schools are meeting their API growth targets is good news (though i'd like to see how many schools at the low end made their targets vs. schools already in the 700s), but adulation should be tempered by an understanding of just what the API is. California's defensiveness is like trying to claim victory after plugging one hole in a swiss cheese boat. The untold tens of thousands of California's kids who still can't read are the real story.
This entry was posted
on Thursday, September 01, 2005 at 3:10 PM.
|
