On April 30, readers talked about high-stakes testing with the authors of a new book on the subject, Collateral Damage: How High-Stakes Testing Corrupts America’s Schools. They are David C. Berliner, the Regents’ professor of education at Arizona State University, in Tempe, and Sharon L. Nichols, an assistant professor of educational psychology at the University of Texas at San Antonio. Below are excerpts from the discussion.
Question: Is “accountability” really necessary for education reform? If so, what are some of the reasons that high-stakes testing is not working to make schools, districts, and teachers more accountable?
Berliner: Accountability is needed for any expenditure of public dollars. But having people give an account—telling what they are doing to educate kids—is not the way we have chosen to view accountability. It’s the narrowing of this concept of accountability to tests that’s one problem, and then we make it worse by narrowing our notion of acceptable data to answers on tests that are cheap to make and score. Both forms of narrowing make the system we have created awful. By putting high stakes on the results of these tests, we have school personnel avoiding real accountability and giving the politicians what they want—high test scores!
Read the full transcript of this chat, “High-Stakes Testing: Helping or Hurting?”
Question: One of the reasons high-stakes testing is a problem is our population’s low sense of morality. So many teachers and administrators are willing to cheat or look the other way. Is there not an ounce of honesty left in our public schools? The No Child Left Behind Act assumes people will be honest. Isn’t that the bottom line with this topic?
Nichols: We think this is crucial. In our book, we talk about how high-stakes testing creates a culture in which cheating is more likely to happen and seems more justified. We have to be careful about condemning our teachers, though. From what we have learned, we suspect that not all acts of cheating are equal. Might it ever be appropriate, for example, to “bend the rules” in the name of helping a student be successful? We don’t condone outright cheating, but, again, we won’t make a blanket statement that adults in schools are cheaters. It’s definitely problematic under high-stakes testing.
Question: What purpose does the testing system serve? Couldn’t a portfolio-assessment system serve the same purpose?
Berliner: The issue, I think, is how important the decision to be made is. If I were a state examiner, I wouldn’t want to base a decision about the knowledge base of a dental hygienist on the portfolio she presented to me. It’s too idiosyncratic and may not cover all the areas I’d want information about. But were I the hygienist’s instructor, I’d want to see the portfolio, and could use it to make sure there was evidence she was progressing well in covering all the things she should be covering. Portfolios, so far, don’t have enough of the psychometric properties needed to make important decisions about kids, schools, or a state’s performance on academic standards. But if there weren’t high stakes attached to these tests, then a portfolio could be used.
Question: In states with high-stakes graduation requirements using state-developed tests, have we seen any indication that there’s been an increase in national testing performance, for example on the SATs and ACTs?
Berliner: Actually, the opposite seems to be true. There are three studies I encountered that show a loss in SAT scores, and I think there was one that showed a loss in ACT scores, as a function of high-stakes testing. This makes some sense. If a state develops standards and tests to match the standards, then it has a more circumscribed curriculum, because its curriculum must reflect those standards and those tests. On the other hand, the SAT and the ACT (especially) have a conception of the common high school curriculum and base their tests on that common, 50-state, typical curriculum. If your state moves into high-stakes testing, you’ve narrowed your state’s curriculum, and it stands to reason that you’ve lost a point here and there on the SATs and the ACTs. That’s what seems to have happened.
Question: High-stakes testing was put in place because we were and are a “nation at risk.” What can be done by the federal and local institutions to make high-stakes testing work? A combination of past and present practices?
Nichols: Well, I think we first need to question whether the risk is real or illusionary. I am not sure we are at risk. But maybe the better way to think about this is not whether we are or are not at risk, but what does it even mean to be at risk in the first place? My co-author, in The Manufactured Crisis, argued that we are not doing as badly as the fearmongers would have us believe, in general. Still, we can always do better. It is just that different schools with different student populations require different solutions. And a significant piece of the puzzle is the economic disparities of our young people’s life circumstances. High-stakes testing is a monolithic answer to a differentiated problem.
And why do we need all students to perform at the same level anyway? Why not create a system that celebrates our students’ diverse interests and talents and allows them to move forward on their own path instead of always defining it for them?
We cannot “make high-stakes testing work” because it does not work. It leads to all types of corruption and distortion of the educational process, and this is not acceptable. We can, however, implement multidimensional practices that minimize the importance of the tests.
Question: Shouldn’t we broaden our discussion of testing and rethink its role in our society, its value, and what it really measures?
Berliner: Yes, you are quite right. We seem to be the world’s leader in numbers of tests that our young people must take, but the world is catching up to our nuttiness. This is all part of the technological age we live in, when it is assumed that the testing technology, like other technologies, works as intended. But it doesn’t. Tests of all types, particularly standardized tests, can offer very useful pieces of information. It’s when too much is made of the scores that everything falls apart. The Europeans make the distinction between testing for instruction (to teach better) and testing of instruction (to see how the kids learned). The former is not done well in the United States, and we rely too much on the latter. Good tests can help us. Bad tests mislead us. We trust too many tests to do jobs they are not psychometrically able to do well at all.
A version of this article appeared in the May 16, 2007 edition of Education Week as Chat Wrap-Up: High-Stakes Testing