18 March 2010

Tests always benefit the test-giver, not the test-taker (even self-assessments)

It occurs to me that tests always benefit the test-giver, and not the test-taker. By tests, I mean pretty much every single assessment method used by teachers everywhere, be they tests, self-assessments, diagnostics, etc.

I could be spending my time right now, for example, on learning a new programming language, writing up a program to test out a game theory model I have in mind, reading research papers, working on my thesis proposal, etc.

Instead, I'm working on final exam test preparation. That means I'm working out some preparation problems that have no relevance to my research, that are not even really educational since I have no solutions to compare and learn from, that are contrived so to direct the test-taker to arrive at a particular solution that is easy for the test-marker to verify.

And why do I have to write an exam for this course? So the course instructor, the test-giver, can write a "letter" on my course transcript indicating my "grade". Think about that again, the test is given for me to write, so that the test-giver can do something (in particular, to scribble down a character on a piece of paper).

Tests benefit the test-giver.

Nothing inherently wrong with that, of course, as the instructor has to somehow write a "fair" letter for my course grade. We just have to recognize the goal, the purpose, of what we do as educators. We as educators demand students to write tests really only to benefit ourselves.

Without tests, students won't know how well they're doing though!!


It's true that students need tests to know how well they're doing, and students can find out how well they're doing by giving themselves tests. That students need to be tested to help them learn doesn't in itself specify who should be giving the test to effect the greatest benefit in their learning. In fact, by being the test-giver themselves, students can assess their own performance. So tests benefit the test-giver, yet again.

And that's the entire point of self-assessments, to have the student test and assess themselves.

Except that self-assessments as practised is a demand placed by the teacher on the student. Students often are not encouraged to, trained to, or naturally come to test themselves or to take up the role of the test-giver. So instead, teachers become test-givers in demanding students perform self-assessments. The teacher still administers the test, often providing critical help in marking the test, which reinforces that the test is teacher given to satisfy the teacher's own goal.

The very psychological conditions that makes putting oneself through a test of ones own making, by design or accident, a personal challenge to learn from are completely missing. The sense of natural rather than artificial or imposed jeopardy is missing. The element of "just in time learning" is completely missing.

Notice the very subtle shift. Look carefully or it'll be missed entirely! The act of completing a test, exam, quiz, etc, are all sets of behaviours. To get a good grade on a test requires engaging in a set of specific behaviours (and we hope those behaviours are indicative of having the requisite set of knowledge).

So when a student gives himself a test, he has to perform the set of behaviours he expects of himself. But when a teacher gives him a test, he has to perform the set of behaviours he thinks the teacher expects of him — even if the test is a self-assessment!

That is fundamentally why tests always benefit the test-giver (assuming a rational test-giver). The test-giver has a goal, wanting something done, and demands of people to perform a set of behaviours (eg, test writing behaviours) that in the end satisfies the test-giver's goal (hopefully. And the test-giver may be the same as the test-taker).

What this means for educators is that if we want the student to benefit from a test that compares how they perform with how they expect themselves to perform, there is essentially no such test possible (self-assessments or otherwise). Precisely because there is no way for a teacher to insert their intentions without taking away from the students' own intentions. It's as though the act of observing affects the observed.

Not all is lost, fortunately, because students can be trained from an early age to test themselves. Once that is habit, they will go on testing themselves without a teacher interfering with their volition. Easier said than done though, because if the tests involves any kind of active teacher assistance (say in marking the test), then the test cease to be of the correct type (because the student would be comparing themselves to what they think the teacher expect rather than what they themselves expect).

The interaction is very subtle, as we are dealing with people's intentions and their expectation of the expectation of what other people intend.

A simple example to wrap up:
People who wants to improve their rock climbing skills will seek out opportunities to climb harder routes, to improve their game and to test what they know. Now imagine a "teacher" comes along and evidently creates a rock face to test a young climber. If the climber fails the climb, he doesn't necessarily have to concede he's bad at climbing and needs to improve, rather he can just concede that the "teacher's" test is too hard, designed to fail the student, and thus he need not improve. This rationalization option simply isn't available if the student picks his own route on a naturally occurring rock face, where if the student fails, the only explanation is that he needs to get better at climbing or pick an easier route.

In that example, the "teacher created" rock face can be easily observed and judged as to whether it is realistic or not compared to a natural rock face. Imagine how much easier it is to rationalize not improving when the "teacher created rock face" is a math or science test, where the student has no clue whether the test is realistic or not in comparison to a naturally occurring math or science test (in fact, I'm not convinced most teachers even know what a naturally occurring math test is in the natural environment of, eg, science research or engineering).

And all that is why tests always benefit the test-giver; why teachers have to be very careful to administer tests that only achieves what they know specifically is the goal; why it's not possible to administer a good-faith self-assessment; and thus why if we want students to be their own test-givers, and to benefit from a good-faith self-assessment, we actually have to train them until it is a habit to give themselves self-assessments; and why those "training self-assessment" tests are notoriously difficult to design.

No comments: