Tuesday, June 21, 2011

Teaching and Testing

This week I received a note from a teaching organization reminding me to call my elected officials to oppose the president's education reform agenda. The organization's flyer emphasized the evils of testing. In particular, testing results might be used to terminate good teachers whose students test poorly.

The heated rhetoric concerning testing is littered with fallacies, from the absurd claims that thousands of "good teachers" might be terminated to accusations that "corporate testing firms" are pushing the tests to make money. I recognize there are political and commercial interests involved in education, but let's take a step back and ask ourselves if testing is always a bad idea.

It's obvious to me why parents and politicians wouldn't understand the union position.

I test students. Most of my colleagues give exams and lab practicums. We assign homework and projects, which are graded. As a profession, then, it is clear that teaching is not automatically opposed to testing — as long as the testing is limited to our students in our classrooms. Testing us or evaluating us based on student tests? We oppose that, even when it seems reasonable to the general public.

I like metrics, assuming the correct metrics are used in a given context.

Unfortunately, the wrong metrics are often applied to situations, resulting in meaningless, misleading, or downright stupid analyses. For example, I had a manager years ago who tracked "daily lines coded" to rate programmers. That is the wrong metric for programmer productivity. In fact, a great programmer strives for condense, fast, accurate code.

But, it would be reasonable to test programmers before hiring them. In fact, I would want programmers to be evaluated before extending a job offer. A test might be debugging a program with common errors. Another test might require the programmer write a basic SQL query extracting data from two tables with a common key. The point is, evaluating an employee's skills and knowledge makes sense, whereas judging the quantity of output regardless of quality is foolish.

If I am teaching a programming course, it is reasonable for my department to wonder how many students can demonstrate improved skills at the end of a semester. For example, I teach a course on design. It's logical to test how many students know the basic HTML/CSS to recreate a theoretical page design with body paragraphs in Times New Roman, indented one em-space, with no space between paragraphs. If the student doesn't know this information, either the student or I failed during the semester. Testing my entire class would reveal if I failed to achieve objectives or if the one student did.

Testing is not a bad metric for measuring knowledge gained. Although we should not test constantly, which takes time from teaching, some testing is valuable.

Contrary to the repeated mantra of education unions, include the national union to which I belong, testing does not necessarily, and certainly should not, require "teaching to the test." I would argue that teaching to the test represents not a failure of testing but a failure of pedagogy.

Not that testing is perfect. But it isn't nearly as flawed as the education establishment wants parents and politicians to believe.

I remember in the late 1980s an education professor asked about high scores on language portions of the SAT. "Those of you with high scores memorized the lists, didn't you? That's the problem with testing," she asserted.

"No," I blurted out, "I read a lot and remember the words from the stories and plays. I recall great writing."

"Well, you're different," was her terse answer.

That is still the attitude in too many education schools. They assume students doing well memorized material in isolated, atomized chunks. I know I couldn't learn that way. When I challenge this, I'm told there's "not enough time" for contextualized learning.

The education establishment wants people to hate testing, because with testing, observation, and parental feedback, there would be a chance to evaluate instructors on their performance. If they can prove testing is flawed, metrics are pointless, administrators are too untrustworthy, etc, they don't have to improve teaching -- they can keep blaming thousands of other variables.

And I am a member of the AFT. They claim testing is a way to weaken the union. Maybe we have some weak members? Maybe our teaching methods are flawed?

Schools have created testing regimes that are absurd. They have allowed the tests to dominate discussions in ways that do make testing, as currently done, a bad metric. But, instead of criticizing testing in general, why not work towards aligning test materials with the subjects taught?

I've met with teachers from across the United States. The best teachers guide students through hands-on projects that are exciting, compelling, and educational. From building model rockets to writing and producing audio dramas, students in these classrooms learn the subject materials that appear on state tests without thinking about the tests on a daily basis.

Innovate, create, explore, and discover. A student creating a simple game on a computer does memorize the underlying computer language and general skills. A student collecting insects or plant specimens for a science class learns about habitats, native species, and much more. Students remember what they do.

The schools that emphasize test-taking skills are short-changing students.

I've had teachers tell me that some students need to learn how to take tests. I understand that some special needs students struggle with tests entirely inappropriate to their limitations. But, again, the solution is not to criticize all testing but to help develop more appropriate metrics for some student populations.

Also, there is a declining return on the statistical value of long tests. We could test students for one to two hours in most subjects and have a relatively accurate measure of performance in any one subject area. Yet, most schools I've visited spend two weeks testing students each year. Two weeks? That's unnecessary for accurate metrics.

I have given 20 question tests and 100 question tests. The grades vary little between the two lengths, so what is the purpose of a longer test? At least in my courses, the longer exam would prove nothing and simply exhaust students. I can ask moderately difficult questions in the shorter exam and determine which students have mastered a topic.

Instead of being vehemently against testing and relying on rhetorical fallacies such as the slippery slope and unsupported protestations of possible anti-teacher bias, unions should embrace reasonable testing practices complemented by sound pedagogy.

Workers in other professions are evaluated by supervisors. Many are tested on a regular basis for continued licensure. To argue that teaching is "special" is to invite more extensive appraisals of our performances. Rhetorical claims of difference start to sound like claims to privilege.

Testing is not going away. What we should do is try to influence the nature of testing to ensure the metrics are chosen wisely and effectively.

No comments:

Post a Comment