Teaching to teach to the test?

US Education Secretary Arne Duncan believes that we need to improve our country's teacher preparation programs - and that we need to use student growth and achievement data to do it.

For several years now, the US Dept of Education has been discussing proposed rules which would require states to rate their colleges of education and like programs. These ratings would also affect eligibility for Federal dollars. As a part of this effort, USED has wanted to include student test scores ("achievement" and/or "growth" data) as part of the ratings - that is test scores of K-12 students who were taught by recent graduates education degree programs. A special committee, comprised of numerous stakeholders, worked nearly a year to come to agreement on new rules, but was unable to do so. So, the Department has gone ahead with it's own ideas of how the people who teach teachers should be evaluated.

The window to comment on these rules closes 2 February 2015; you can read the details and submit your own comments to Federal regulators via this link. For those who might struggle to find the right words, we reprint our own comments, submitted earlier today, below. We hope that we were able to articulate just a few of the many objections to using standardized test scores in this manner.

On behalf of public school parents across the state of Michigan, we would like to comment on the proposed rule changes with regard to the evaluation of teacher preparation programs. Specifically, we are deeply disturbed by the inclusion of student achievement and student growth measures as defined in proposed subsection 612.2(d).

There is no question that the professional quality of new teachers is very important to their own success, that of their students, and the overall success of our schools. (This importance should not overshadow the need for continuous education, support and training throughout a teacher's professional career.) However, we are very concerned about the use of measures of student achievement and growth when these measures are based on narrow, standardized tests in a small handful of subjects. We have objected to their use in the evaluation of teachers, schools, and districts, and for the same reasons we oppose their use in the evaluation of teacher preparation programs.

Impact of poverty
The evidence is clear that student income level is a primary determinant of standardized test scores of all kinds. Students living in poverty, in particular, tend to have lower scores on such tests, regardless of other factors such as ethnicity or geography. In Michigan, for example, the percentage of students in a school building living in poverty (eligible for free lunch) alone explains between 50% and 60% of the variation in building mean scale scores on the Michigan state tests. [See http://www.mipfs.org/node/181]

Likewise, it seems reasonable to expect that poverty - insecure home situation, insufficient nutrition and medical care, homelessness, and other conditions endured by children in poverty - will also affect a student's ability to "grow" during the school year. These problems are not left behind at the schoolhouse door.

Efforts to use student test scores or test score growth to calculate "value added measures" of teacher effectiveness are dogged by the same problems. Academic research into the construction of VAM models frequently uses measures - such as actual permanent family income, housing stability, parental education level - which are not available to school districts. The research shows that these out-of-school factors account for the vast majority of a student's achievement as defined by test scores. Where school districts and states attempt to use these models without including these important variables (a yes/no indicator for free or reduced lunch status does not substitute for knowing actual permanent family income or wealth), the results are subject to tremendous error and bias that would make any conclusions highly suspect.

Using flawed measures of student "achievement" to evaluate teacher preparation programs, when these measures are clearly subject to bias and large errors, would have the effect of penalizing programs which tend to send their graduates to communities with the most needy students - even if these graduates have in fact received training that is of high quality. It would, in fact, create perverse incentives for colleges of education to avoid placing their graduates in high-poverty schools and districts as much as possible. Surely this is not the intent of the proposed regulations.

Narrow scope
Standardized tests in two or three narrow subjects may indeed be a useful diagnostic tool, once other confounding factors have been adequately accounted for. However, they do not and cannot assess the full range of learning which parents expect their schools to provide. While better test scores may be a secondary consequence of a quality education (see the statement, which we joined, here: http://parentsacrossamerica.org/quality-education/), it is also possible to prepare students to do well on standardized tests without providing them with the reasoning, collaborative and creative skills a quality education imparts.

As parents, we insist that prospective teachers be trained to offer our children a true quality education. As citizens, we know that the future of our communities depends on it. Evaluating teacher preparation programs on measures that do not reflect real educational quality will again provide perverse incentives - in this case, to train teachers who will focus on test-taking skills and narrow acquisition of certain facts. This is not what we want for our schools or our children.

We welcome efforts to improve the quality of teacher education, including measures to support mentoring programs and other post-degree training. We also welcome Federal support and funding of such efforts. But Federal regulations must be based on the fact that a true quality education is broad in scope, includes a wide range of skills and attributes which are not easily tested, and is best assessed over the long term. The current draft regulations do not meet that standard.