Interesting Statistics Problem.

[VTBoy]

Suppose in 12 indendent tosses of a coin 9 heads and 3 tails are observed.
Test the null hypothesis theta = 1/2 versus the alternative hypothesis theta > 1/2, where theta is the true probability of heads.

Given this information and only this information test the hypothesis.

[hhEb09'1]

Sounds like homework :)

Can you wait a few days?

[VTBoy]

Sounds like homework :)

Can you wait a few days?

Oh no this isn't homework.



It is to show that most people will make an assumption when they see this problem, on how to solve it. When that assumption is not valid. This problem is in one of my advance stat books.

[hhEb09'1]

It is to show that most people will make an assumption when they see this problem, on how to solve it. When that assumption is not valid. What assumption do they make?

Is the point of the problem in the textbook to point out the invalid assumption? Which text is it?

[VTBoy]

What assumption do they make?

Is the point of the problem in the textbook to point out the invalid assumption? Which text is it?


In order to do the problem you need to determin the distribution, and most people would assume binomial distribution. But that is assuming too much. The distribution could infact be negative binomial. It depends on why the experiment was stoped, which isn't stated.

The book is Bayes and Emperical Bayes Methods. It is a section that is critical of classical statistical inference.

[hhEb09'1]

The book is Bayes and Emperical Bayes Methods. It is a section that is critical of classical statistical inference.Of course Bayes would say that! :)