I like John Kulig's suggestions. In fact, I do them all. As for the problem
with runs of the same answer over several questions, having too many of one
answer, etc., this is what I do: As I write test questions, the first will
have A as the correct answer, the second will have B, the third will have
C, and the last will have D. I then start over again with A. When I am
done, I randomize the order when I print the test (by hitting the
random-sequence key). This way I have an equal number of each answer on the
final test, and they are in a random order (according to the randomization
function of the testing program). Of course, this doesn't work if you use
test-bank questions provided by the publisher. But I don't like these test
banks, so I usually write my own MC questions.
Jeff
"John W. Kulig" wrote:
> Tipsters:
>
> The thread on "C" thought me thinking about ways to remove non-valid
> "cues" on multiple choice exams. What follows is a short list of things
> I try to avoid. I'd be curious to know if others have worked on this
> issue as well.
>
> Avoid style mismatch: E.g. "The cell body of a neuron is called a
> (1) soma (2) axon (3) dendrite .." a guesser can eliminate (2) because
> "a axon" Therefore use " a/an".
> Equate length of answers: Make sure they are all of similiar length
> (at very least, make sure length of answer is uncorrelated with truth
> value).
> Avoid obvious wrong answers. "Most psychologists agree the best way
> to raise children is to (1) beat them every day whether they deserve it
> or not (2) let them write on the walls (3) establish rules, but, be
> flexible ..". Even people who haven't purchased the textbook can
> eliminate (1) and (2). I have seen choices almost as absurd on others'
> exams.
> Avoid aving too many of any one letter correct.
> Avoid the "gambler's fallacy" - resist the temptation to avoid
> repetition. The exact sequence a-a-a is just as likely as b-c-a (even
> though the latter _looks_ more random). It's easy to pull a-b-c-d slips
> from a hat.
> Break expectations students have developed. If you have sereral
> answers in a row that have the same possible answers, it's OK to have
> the same answer more than once. E.g. "Teaching rats to press levers to
> obtain food rewad is an example of (1) operant conditioning (2)
> classical conditioning (3) habituation". Most students will assume that
> if (1) is correct, it can be eliminated on the next question which might
> read "Teaching autistic children to have manners with verbal rewards is
> an exampe of ..". and so forth. I once did an exam where the answers
> were (1) hypothalamus (2) thalamus (3) corpus callosum (4) cerebellum. I
> had "hypotlamus" correct 3 times in succession.
>
> In the limiting case, we'd expect "chance" to be 25% for a 4 choice
> exam, and we could defend a curve that had D >= 25%. That seems
> outrageous, but not when all non-valid cues are removed.
>
> --
> ---------------------------------------------------------------
> John W. Kulig [EMAIL PROTECTED]
> Department of Psychology http://oz.plymouth.edu
> Plymouth State College tel: (603) 535-2468
> Plymouth NH USA 03264 fax: (603) 535-2412
> ---------------------------------------------------------------
> "The only rational way of educating is to be an example - if
> one can't help it, a warning example." A. Einstein, 1934.
--
Jeffry P. Ricker, Ph.D. Office Phone: (480) 423-6213
9000 E. Chaparral Rd. FAX Number: (480) 423-6298
Psychology Department [EMAIL PROTECTED]
Scottsdale Community College
Scottsdale, AZ 85256-2626
"The truth is rare and never simple."
Oscar Wilde
"Science must begin with myths and with the criticism of myths"
Karl Popper
"If you want to learn new things, you should try reading old books."
Richard Cytowic
