Re: Grading Intelligence Tests

From: David Bandel (dwb1729_at_yahoo.com)
Date: 06/27/04 

Date: 26 Jun 2004 23:46:26 -0700

Ioannis <morpheus@olympus.mons> wrote in message news:<1088278607.950085@athnrd02.forthnet.gr>...
> Assuming one constructs a very hard intelligence test with 50 questions,
> how should such test be graded?
>
> Specifically, I've constructed a very hard 50 question test and I was
> wondering if I should assume a normal distribution, so I could "assign"
> an IQ of 100 to 25/50.
>
> This wouldn't be such an interesting question, apart from the fact that
> although the median should (?) be around 25/50, nobody (from a sample of
> 30 people) got any higher than 17/50.
>
> What's wrong with my assumptions?
>
> Thanks much,

how many questions they get correct is arbitrary. what you want to do
is assign an IQ of 100 to the average amount of questions gotten
right.. then from there apply the bell curve for distribution etc.
25/50 means nothing..

Quantcast