Friday, December 7, 2007

Practical advice on writing objective exams

Hello everyone,

As the semester comes to a close at Garfield, you should be planning for the final exams in your classes.

Generally, final exams are summative assessments, though first semester finals can be used to guide instruction in many of your classes' second semesters. I want to think in summative terms, however, and I want to consider only written exams.

Let's put grading aside, and consider only how to write a final exam as a summative assessment. There are a few rules that you MUST keep in mind:

1. You cannot hold students accountable for something they haven't been required to learn. Avoid including questions on your final that reference anything that wasn't covered in some explicit way in class.

2. Your questions must be clear, understandable, and unambiguous. Do not use trick questions; do not hold students fully accountable for ambiguous questions.

3. Written exams used as summative assessments cannot measure everything a student has learned if your subject teaches both skills and concepts, unless the skills you are testing are reading, writing, or calculation. Practical skills like performing music, playing a sport, or using equipment must be assessed using another kind of test.

Keep in mind the general rule that essay tests are easy to write but hard to correct. Objective tests (multiple choice, true and false, matching) are easy to correct but hard to write. As a former English teacher, I prefer essay tests, but there is rarely time to correct them (unless you're very fast) at the end of the semester. Having said all that, I want to give a few suggestions on objective tests.

MULTIPLE CHOICE

I think this is the best objective method for measuring knowledge, but you need to be careful how you write them or your assessment will be skewed.

You should give students four or five choices. Make sure that the possible answers are spread randomly over the full range of possibilities. (Letter "B" is the most commonly used answer slot; consciously make sure that "A," "C," and "D" are used equally as much.

Make sure that the answers are unambiguous. There should only be one correct answer. If a student points out that you've mis-written a question or the answers, be prepared to give credit for multiple answers.

Most students can easily answer 100 questions in an hour.

TRUE OR FALSE

These are easy to write and correct, but do not give a very good assessment because they are so easy to guess on. I don't think it's fair to make the questions hard to understand to compensate for the ease of answering them. Mixing a few true/false questions in a lengthy test, however, is like giving the kids a break.

100 questions in 1/2 hour.

CLOZE

This is the sort of question that makes a statement, leaving out a key vocabulary word. Students must be able to put the correct word in a blank. It is very easy to get too ambiguous with Close tests. Make sure that only one word could possibly go in the blank. "Marie Curie is credited with the discovery of the radioactive substance named ______." This sort of test is good for cold recall, but teachers often get caught up in what they WANT to go in the blank, and forget what COULD go in the blank, which is frustrating for students.

100 questions in 75 minutes.

MATCHING

This requires students to match one thing in one column, with a similar thing in another column. Having students match more than ten pairs of words or concepts is usually difficult and confusing, while matching five pairs gives just as fair an assessment, if they really know the material. Having words that could be the answers for more than one question, or having words that don't match with anything provides a more accurate assessment than having matchups that allow the process of elimination to dictate thinking.

100 questions in 50-90 minutes.

I hope this is useful. Remember, you're trying to assess them, not stump them. It's a test, not a trick.

Jeff Combe

No comments: