Wednesday, August 4, 2010

More on Testing

This may be familiar ground, but an article by Richard Rothstein in American School Board Journal made me think it's worth repeating, particularly in light of Pennsylvania’s ongoing march towards state-wide “Keystone exams”.

An argument you occasionally hear is that schools should be "run more like a business". But as Rothstein pointed out, performance pay in the private sector is seldom based on strictly quantifiable measures of how much is produced.

In fact, he gave a number of examples of how perverse incentives are sometimes built into systems.

For example, if colleges are ranked according to how "selective" they are, they'll work at getting more students to apply, just so they can turn them down. 

A classic example: factories in the Soviet Union were rewarded/punished according to how many shoes they produced. But because material was in short supply, they produced a lot of very small shoes - in sizes nobody wanted.

The consequence for education: if schools are measured strictly, and with important consequences, by the percentage of their students who reach a certain threshold (such as "proficient"), they will be tempted to devote a disproportionate share of their resources to the students just below the threshold, at the risk of ignoring everyone else.

Here’s another: in many high schools, "college-prep" English courses are designed to help students do well on the standardized tests (with their emphasis on grammar, which is easily testable) that will help them get into college - but doesn't provide the skills that will make them successful in a freshman-year English course.

What's the moral of the story?  In part: not everything that is important can be measured, and not everything that can be measured is important.

When it comes to assessment, the United States is an international outlier. As Stanford University’s Linda Darling-Hammond has shown, many nations with better and more equitable educational outcomes test far less than we do. They typically test just one to three times before high school graduation, and use multiple-choice questions sparingly, if at all. Excessive testing wastes resources and fosters the use of cheap, low-level tests, while adding high stakes and potentially narrowing the curriculum. The results provide little instructional value to students, teachers, schools, or districts.
Arne Duncan is beginning to suggest that he understands this: “Only by moving beyond basic skills and bubble tests can children develop the critical thinking skills that will one day give them the ability to compete successfully in our increasingly global, increasingly competitive international economy." It still remains to be seen whether the policy will match the rhetoric.

So here's the question: Is it possible to assess critical thinking, innovation, and communication skills on standardized tests?  If not, why are we going down this road?