The New York State Education Department has released the data from the standardized tests administered to elementary students at the end of this past academic year.  The opt-out movement scored a much bigger victory (if victory was its goal) than I ever would have imagined.  Some students opted out for valid health and other reasons, but a whopping twenty percent of students refused to take the tests without any valid excuse.  Presumably they objected based on principle.

A five or so percent opt out would have done little to upset the status quo, but one of this size has enormous ramifications.  First, as a condition of receiving federal education money, the Elementary and Secondary Education Act (No Child Left Behind) mandates that 95% of students take the test.  The Secretary of Education has the power to waive a number of requirements, but the Secretary cannot waive this requirement. Congress wanted this one to stick.  As analyzed in an earlier post, this poses a real quandary.  The purpose of the provision was to make sure that schools did not conveniently exempt their weakest students from the test to push up their pass rate.  But when students simply refuse to take the test, holding the school accountable seems unfair and contrary to the purpose of the Act.  The Secretary could take a page from the reasoning of King v. Burwell (the recent Affordable Care Act decision offering a creative reading of the Act to save individual tax credits) and ignore the statutory language and violation.  But absent creative reasoning, New York is in violation it cannot escape.

Second, those who opted out did not represent the overall student body.  The data shows that those who opted out were disproportionately  white, middle income, and speak English as a first language.  They, however, were also less likely to have achieved proficiency on the prior year's exam.  This presents an interesting mix.  My earlier analysis did not assume this last one--that they were low scorers.  Thus, I feared that the movement could fuel more segregation by skewing testing down in certain areas, as high scorers opted out.  It seems, however, that the result will be to drive average test scores up in predominantly white middle income schools and districts.  The net result, however, is the same.  The size of the achievement gap between whites and students of color could be inflated, as well as the gap between the overall achievement in advantaged versus disadvantaged districts.  The result of such a gap could further incentivize parents to not just opt out of tests but opt out of certain schools and districts.   If ever there were a perverse incentive of No Child Left Behind, this would be it.

Third, opt-outs represent missing data.  When missing data is small and random, it can more easily be accounted for, but here that would seem impossible, at least, the way that data is being put to use.  States, including New York, have developed complex statistical models to try to attribute student achievement on standardized tests to a teacher's effectiveness.  As demonstrated in Part II of this article, these models are of highly questionable validity as a general principle.  One teacher in New York has already sued.  But when twenty percent of the student population opts out and they are non-representative, the statistical analysis may be pointless.  It is hard to imagine Humpty-Dumpty can be put back together again with some many key pieces missing.  New York's value added method may still spit out scores for teachers, but I would expect the unreliability of those scores to be through the roof for many teachers.  What then will schools and districts do?  That takes us to the next problem.

Fourth, one of the central pillars and conditions of receiving a waiver from No Child Left Behind was to commit to evaluating teachers based on students' scores and making personnel decisions based on them.  I demonstrate here that such a condition was illegal.  But until a court agrees or Congress passes a newly reauthorized Elementary and Secondary Education Act, Secretary Duncan is in charge.  This means New York must either press on as though nothing is wrong with the results or Duncan will have to back off of New York's waiver requirements.  If the state presses on, it only digs a deeper hole for itself in defending against the due process challenge to its value added method of evaluating teachers.  If Duncan flinches, he opens the door to more states (and teacher unions) insisting on more leniency on the same measure.  He already allowed for states to delay implementation of the systems once.  This time, however, a fundamental flaw, rather than just implementation problems, is being exposed.

Given that, who is up for ESEA reauthorization now so we can avoid this mess?  Well, that is a whole set of different problems. See here.