And lo On Fri, 12 Feb 2010 21:11:07 -0000, Jim Charter <jrc### [at] msncom>  
did spake thusly:

> I am marking some 50-question, 4-choice, multiple choice tests. (About  
> 20 of them.)  The scores are noticeably low, 50%, 64%, 38%, 44%, 52%,...
> I begin to worry that the marking key I am using may be mismatched with  
> the version of the test I am marking.  After about nine papers marked,  
> just as I am about to review the key against the actual questions,...I  
> get a score of 84%.  I am immediately and intuitively certain the key is  
>   the correct one and that this test group is just unusually low  
> scoring.  Is there any science that discusses this sort of phenomena?

Okay think what a number of students/result graph would look like if there  
were no questions, just random ticks of the box. You'd expect a peak  
around the 25% mark tapering off on either side.

Now if there are questions and the people taking the exam are expected to  
know the answers you'd expect the peak to move to the right. Just like  
asking a mathematics class -

What is 1+1?
a)1
b)2
c)3
d)4

And similar easy questions.

Now imagine a misprint so the question reads - What is 1*1? What would you  
expect to happen to the graph with the same class? The peak would move to  
the left. This would also be the equivalent of having the wrong answer  
grid.

So taking the results you've given I'm just guessing you've got a  
low-scoring class; or the 'wrong' answer grid you're using is very similar  
to the 'correct' one.

-- 
Phil Cook

--
I once tried to be apathetic, but I just couldn't be bothered
http://flipc.blogspot.com

Post a reply to this message