Slashdot Mirror

According to people like Diane Ravitch, who was assistant commissioner of education under both GHW Bush and Bill Clinton, the data is still too noisy.

They can't correct for socioeconomic makeup because schools have no way of knowing family income unless the kids are in a free lunch program.

Ravitch has a PhD, she spent her life studying the data. She started out believing in high-stakes testing, and now she says she had to change her mind because the data doesn't support it.

See
http://www.nybooks.com/blogs/nyrblog/2012/feb/21/no-student-left-untested/
No Student Left Untested
Diane Ravitch

http://aera.net/uploadedFiles/Gov_Relations/GettingTeacherEvaluationRightBackgroundPaper(1).pdf as her supporting data.

If the LA tests found that there was no correlation between the judgement of the administrators and the results of the tests, that doesn't mean the administrators were wrong. The tests could have been wrong. It could have been the tests that were no better than random, which is what seems to have happened in NYC.

I'd like to see anything that shows the LA tests were scientifically valid.

BTW, one teacher in LA killed himself after his low test results were posted.

Your understanding of statistics and scientific validity is naive.

The tests don't work. They don't measure achievement with enough accuracy to use them for anything more than a rough guide. People are assuming that the tests are valid, when they're not.

This is explained in the teachers' newspaper ad linked in TFA, and also here:

http://www.nybooks.com/blogs/nyrblog/2012/feb/21/no-student-left-untested/
No Student Left Untested
Diane Ravitch

http://aera.net/uploadedFiles/Gov_Relations/GettingTeacherEvaluationRightBackgroundPaper(1).pdf

Diane Ravitch says the same thing you did. The National Academy of Education says the same thing you did. Every scientific review of these teacher evaluation systems says the same thing you did.

If you want to be fair to the teachers -- which most people are not -- before you fire them, you should read what they have to say in their own defense in the ad linked in TFA http://www.uft.org/files/attachments/uft-teacher-data-reports-formula-ad.pdf . They say that the tests have a huge margin of error. Teachers in the top 50% could be rated in the bottom 50%, and vice versa, simply because of the error ranges in the testing system that have nothing to do with how well their students learn.

If you want to conceptualize it, consider this: A good teacher in a top school with good students might have a class in which the average grade is 98%. Students like that have nowhere to go on a standardized test. They're already at the top. When they run the formula on teachers like that, they get bad evaluations because their students aren't improving.

Here's what Diane Ravitch said. I recommend the entire article for those who want to be fair to the teachers and are interested in the facts:

http://www.nybooks.com/blogs/nyrblog/2012/feb/21/no-student-left-untested/
No Student Left Untested
Diane Ravitch

New York’s education officials are obsessed with test scores. The state wants to find and fire the teachers who aren’t able to produce higher test scores year after year. But most testing experts believe that the methods for calculating teachers’ assumed “value-added” qualities—that is, their abilities to produce higher test scores year after year—are inaccurate, unstable, and unreliable. Teachers in affluent suburbs are likelier to get higher value-added scores than teachers of students with disabilities, students learning English, and students from extreme poverty. All too often, the rise or fall of test scores reflects the composition of the classroom and factors beyond the teachers’ control, not the quality of the teacher. A teacher who is rated effective one year may well be ineffective the next year, depending on which students are assigned to his or her class.

and she cites the NAE study http://aera.net/uploadedFiles/Gov_Relations/GettingTeacherEvaluationRightBackgroundPaper(1).pdf as her supporting data.

You're wrong. That's the problem. The tests can't measure performance above expectations because there is no way to figure out what the expectations should be. The evaluations are scientifically invalid.

The UFT ad in TFA makes that argument. If you want to be fair to the teachers before you fire them, you ought to at least read what they say in their defense.

http://www.nybooks.com/blogs/nyrblog/2012/feb/21/no-student-left-untested/
No Student Left Untested
Diane Ravitch

New York’s education officials are obsessed with test scores. The state wants to find and fire the teachers who aren’t able to produce higher test scores year after year. But most testing experts believe that the methods for calculating teachers’ assumed “value-added” qualities—that is, their abilities to produce higher test scores year after year—are inaccurate, unstable, and unreliable. Teachers in affluent suburbs are likelier to get higher value-added scores than teachers of students with disabilities, students learning English, and students from extreme poverty. All too often, the rise or fall of test scores reflects the composition of the classroom and factors beyond the teachers’ control, not the quality of the teacher. A teacher who is rated effective one year may well be ineffective the next year, depending on which students are assigned to his or her class.

and she cites the NAE study http://aera.net/uploadedFiles/Gov_Relations/GettingTeacherEvaluationRightBackgroundPaper(1).pdf as her supporting data.

Read this article. Ravitch was assistant secretary of education under both GHW Bush and Bill Clinton. She has a PhD, and she understands the statistics and data much better than you or I ever will. She started out believing in these educational reforms, particularly charter schools, and she said that, when the data came in, the reforms didn't work.

http://www.nybooks.com/blogs/nyrblog/2012/feb/21/no-student-left-untested/
No Student Left Untested
Diane Ravitch

Most significantly, in terms of what we were discussing, she says:

New York’s education officials are obsessed with test scores. The state wants to find and fire the teachers who aren’t able to produce higher test scores year after year. But most testing experts believe that the methods for calculating teachers’ assumed “value-added” qualities—that is, their abilities to produce higher test scores year after year—are inaccurate, unstable, and unreliable. Teachers in affluent suburbs are likelier to get higher value-added scores than teachers of students with disabilities, students learning English, and students from extreme poverty. All too often, the rise or fall of test scores reflects the composition of the classroom and factors beyond the teachers’ control, not the quality of the teacher. A teacher who is rated effective one year may well be ineffective the next year, depending on which students are assigned to his or her class.

and she cites the National Academy of Education study http://aera.net/uploadedFiles/Gov_Relations/GettingTeacherEvaluationRightBackgroundPaper(1).pdf as her supporting data.

I don't have a PhD, and I don't understand statistics and the data as well as Ravitch does, so I can't help you understand this any more. But this is what the peer-reviewed literature consistently says. This is what I read in Science magazine. This is what the National Academy of Education says. Who are you going to believe, Mayor Bloomberg or the expert panel of the National Academy of Education?

Using student test scores to measure teacher's teaching ability doesn't work. The evidence is against it.

Sorry, I was out yesterday, I hope you get this reply.

These people *do* consider themselves "social scientists," but many use artistic techniques for arriving at their conclusions. Peruse the American Education Research Association's Annual Meeting catalog sometime to get a hint at what these people consider to be science.

http://aera.net/meetings/Default.aspx?menu_id=342&id=4850