"John Knight" <johnknight at usa.com> wrote in message
news:_bE%8.17891$Fq6.2152919 at news2.west.cox.net...
>> "Parse Tree" <parsetree at hotmail.com> wrote in message
> news:EYe%8.9721$sb5.694385 at news20.bellglobal.com...> > "John Knight" <johnknight at usa.com> wrote in message
> > news:Ce7%8.15841$Fq6.1704086 at news2.west.cox.net...> > >
> > > If the test designers equally distributed the correct answers over A.,
> B.,
> > > C., and D. (which they did), then the correct answer will be selected
> 25%
> > of
> > > the time (if all students just guessed at random).
> >
> > Random guesses are independent of what the actual answer is. That's why
> > thy're called 'guesses'.
>> This is true. The argument had been made that a non-even distribution of
> how the correct answers were assigned to different letters, A), B), C),
and
> D) would bias the results, but the reality is that a random selection
> cancels this out as a possible factor.
Yes. Whoever made this argument is wrong.
> Also, the 3% standard error is too big for this distribution. Where 25%
of
> the students answered A) by chance, the standard error is closer to 25% x
3%
> = .75%, rather than 3%.
I wouldn't know, since I haven't seen any of the actual data gathering
methodology.
