Rick & Herb:
A simple way to correct for chance on true/false or MC is to use
the formula R - [W/(n-1)] where R = # correct, W = # incorrect and (n-1)
= # of alternatives to chose from. On a true/false (n-1) = 1 because
there are two alternatives (true and false). On a true/false test, then,
a score of 5 correct out of 10 would convert to 5 - 5/1 = 0; 10 correct
out of 10 would convert to 10 - 0/1 = 10. 8 correct out of 10 would
convert to 8 - 2/1 = 6. The same with MC except (n-1) = 3 for a 4 choice
question. It simply drops down the observed # correct by W/(n-1) which
is an estimate of the # obtainable by chance.
This rescales the # so that chance performance = 0 and perfect = maximum
# of questions - logical. You can then convert to %. But, of course, if
you travel down this road, logical as it may be, you'll encounter a lot
of interesting issues. Is passing a course anything above chance? Then
set your D cutoff at 25% raw score (or 0 corrected). But is chance
REALLY at (n-1)%? Most exams have questions that can be answered by
someone who knows nothing.
The biggest obstacle, I think, is that it'll take a lot of explaining to
students, some will never get it. And if they fail your class they will
blame the grading scheme - and its creator! Where I went to graduate
school (if memory serves) intro psych grades were expressed as T scores
(m=50 s=10) to deal with the fact that on very tough exams, 60% raw
scores was a decent grade. However, it was part of a shared culture, and
the undergraduates were a pretty elite group who would cope with this. I
have played with Z and T scores (uncorrected) but I would never do it
again in my current environment. I also think the chance issue is more a
problem for "criterion" referenced tests - like a driving test or
qualifying exam, rather than a liberal-arts course, where "relative
standing" makes most sense regardless of where true chance lies.
Interesting issue, however.
============================================
John W. Kulig
Professor of Psychology
Plymouth State College
Plymouth NH 03264
============================================
"Nothing will benefit human health and increase chances for survival of
life on Earth as much as the evolution to a vegetarian diet" - Albert
Einstein
> -----Original Message-----
> From: Herb Coleman [mailto:[EMAIL PROTECTED]
> Sent: Monday, December 13, 2004 1:21 PM
> To: Teaching in the Psychological Sciences
> Subject: Re: Correcting for chance
>
>
> > Subject: Correcting for chance grading on multiple choice quizzes
> > From: "Rick Froman" <[EMAIL PROTECTED]>
> > Date: Sat, 11 Dec 2004 16:14:05 -0600
> > X-Message-Number: 3
> >
> > ...
> > Has anyone ever considered or implemented a grading procedure for
> > multiple choice tests where you only start counting points earned
above
> > chance? For example, in the case of my 15 item three-choice
quizzes, 6
> > correct would be one point, 7 would be two points, etc. for a total
> > possible of 10 points. Any thoughts on doing this?
> >
> > Rick
> >
> > Dr. Rick Froman
>
>
> This year I started requiring my students to get at least 50% to get
> credit for their multiple choice items. I give 50 item tests so
someone
> score 0-24 would get 0. If the score 25-50 they get that score (for
> that half of the test). I guess I set my threshold higher than chance.
>
> --
>
> Herb Coleman
> Instructional Technology Manager
> Adjunct Psychology Professor
> ACC/AFT Local 6249
> Austin Community College
> [EMAIL PROTECTED]
> 512-223-3076
> *************************
> * There is still hope *
> *************************
>
>
>
> ---
> You are currently subscribed to tips as: [EMAIL PROTECTED]
> To unsubscribe send a blank email to
[EMAIL PROTECTED]
---
You are currently subscribed to tips as: [email protected]
To unsubscribe send a blank email to [EMAIL PROTECTED]
