"John Knight" <johnknight at usa.com> wrote in message
news:n3W19.49073$Fq6.4310182 at news2.west.cox.net...
> "Parse Tree" <parsetree at hotmail.com> wrote in message news:I1G19.1990
> > > Believe it or not, Brian, the "liberals" are still arguing over what
the
> > > distribution would be if test takers just randomly guessed at the
> answers
> > to
> > > a four choice multiple choice question.
> > >
> > > The distribution over 10,000 test takers will be as follows:
> > >
> > > A) 25%, plus or minus 0.75%
> > >
> > > B) 25%, plus or minus 0.75%
> > >
> > > C) 25%, plus or minus 0.75%
> > >
> > > D) 25%, plus or minus 0.75%
> > >
> > > Without wasting any time with all their silly and erroneous
suppositions
> > and
> > > assumptions (the kind of thing they must have gone through when they
and
> > or
> > > their cohorts answered lower than if they'd just guessed on ONE THIRD
of
> > the
> > > questions), do you agree or disagree that this would be the
> distribution?
> >
> > This WOULD PROBABLY be the distribution. You keep using probabilities
> while
> > trying to speak with certainty. You can't do that.
>> Ah, are we making progress?
>> "This WOULD PROBABLY be the distribution"?
>> Are you now making an about-face?
Nope. I've said this the whole time, because I have a firm grasp of
probability.
> You actually DO agree now that this would be the outcome?
No. I agree that it PROBABLY would be the outcome. There is no certainty.
> It is ONLY by using probability and statistics that we could arrive at the
> conclusion that "This WOULD PROBABLY be the distribution". There's no
other
> way to know.
Yes, only by using probability could we determine if something is probable
or not.
> And the fact is that probability and statistics tells us that this WILL be
> the distribution, which means all your feminazi cohorts who proclaim from
> the rooftops that "we just don't have enough information" or "this is
> ambiguous" are DEAD WRONG, and mathematics is DEAD RIGHT ON.
No, they don't. Probability really doesn't deal with certainties. It deals
with uncertain values, for the most part.
It is PROBABLE that the distribution would be uniform.
> The larger the sample size, the smaller the deviation from 25%. TIMSS was
a
> large enough sample size that a 3% standard error is conservative.
I don't know the size of the sample. Was it 10 000?
> So if 10,000 students just guess at a 4 part multiple choice question
which
> they have no clue what the answer is, between 24.25% and 25.75% of them
will
> accidentally get the correct answer.
Will PROBABLY get the correct answer. It is possible that 0% could get the
correct answer.
> If 20% of them select the correct answer, then we know they didn't just
> guess randomly, because they scored 5% lower than if they'd just guessed.
> This means they have to know something and either:
Nope. 20% guessing a particular answer is certainly in the realm of
possibility.
> 1) Knew the correct answer but intentionally selected the wrong answer,
or
>> 2) What they "knew" was wrong.
>> What's your opinion of why American 12th grade girls scored lower than if
> they'd just guessed on ONE THIRD of the questions?
>http://christianparty.net/timssphysics.htm
If they guessed and had an even distribution, unto themselves.
You know that an individual person could quite easily guess on every
question and get 0. So essentially, the guessing stat is useful only for
the aggregate. And it is only a probability measure.
