Are there any writings/researches out there concerning the topic of "Multiple-Choice Tests" in matters of statistics and probability (i'm specially interested in the psychometric exam).

thanks
and excuse me if it's not the right place to ask

I know that CFA Institute (Chartered Financial Analyst) has recently changed the nature of their multiple-choice exams because of research in this area. I don't have the references here at home, but when I get to the office I'll hunt around.

The short answer is, "Yes, there is research in this area."

There was a similar question to yours at Yahoo Answers. It's not a complete examination of the topic, but rather a good basic introduction.

You might want to search for information about binomial distributions for better answers to your query.

Wow thanks, that's great news.

S2000magician, I'd love to hear what you came up with, thank you.

One gripe I've had for years is with what we called the multiple multiple choice format. I've pointed out the weakness of this testing method to teachers and professors yet they continue to generate exams with this flawed technique. That's the one where all are correct or none are correct or A and C are correct or B and D are correct or only D is correct or A, B, and C are correct etc. If the answer is all are correct, you penalize the student the same amount if he thinks all are incorrect compared to the one who thinks A, B, and C are correct. There's no partial credit. This format throws away useful data. If they convert one of their questions to a series of 4 or 5 true-false questions, they'll get a more accurate assessment of what the student knows. Yet this format seems to be pervasive throughout colleges and professional schools.

Larry

There was one notorious course where I did my undergraduate studies.
There are 20 multiple choice questions in the exam - you get 5 points for correct answer and -2 for a wrong one.
The answers are always very very close to one another.
This course is known to fail 90% of the students so they had to factor the end result. Usually what was done was a square root factor. Basically you take the square root of the result and multiply by 10 and that is your new score. so if you got 81 out of a 100 you would end up with 90. If you got 36 of a 100 you would end up with 60.

one time we looked at the final scores and some results were given as complex numbers...

I'm just fooling myself here - there can never be a pattern, there can never be any rule which'll tell you the excat location of the ansers.

I just figured, if we could beat the odds in blackjack, then theres gotta be some genius out there who figured a way to do so here as well.

Now don't get me wrong, it's not that I'm asking for my personal benefits here and now, although I do need to take a test of the like soon, I'm tearing my *** for it. I was just interested to know whether there was someone who "cracked the code"

Nevermind that, I'd still love to hear more about the subject, whatever you got.

cheers

I work for ACT (American College Testing) and there's a TON of information and resources on our web site: http://www.act.org/

The two references from CFA Institute are:

Writing Test Items to Evaluate Higher Order Thinking by Thomas M. Haladyna (Allyn and Bacon)

Developing and Validating Multiple-Choice Test Items (same author)