Re: Jarque-Bera test: confidence intervals for normal data
- From: Jack Tomsky <jtomsky@xxxxxxxxxxxxx>
- Date: Fri, 16 Mar 2007 22:21:05 EDT
***How can your procedure be correct if the
acceptance region according to your definition is
undefined? What kind of procedure is it if you can
never accept the null hypothesis? Jack***
My response
At that significance level there are any procedure
solving the problem. All persons are aware.
The current conclusion is:
_____at the 6,25 % level we do not reject
H0 : p<=0.5 if we observe X<=4,
However if X=5 H0 is rejected. .
__________licas (Luis A. Afonso)
In your example, N=4. So under your acceptance rule, since X is always less than or equal to 4, the null hypothesis is always accepted.
If you really meant to accept Ho if X <=3 and reject if X=4, then you are now correctly defining b, not as the x for which F(x) = 1-alpha as you originally defined b, but as the smallest x such that F(b) >= 1-alpha. This makes the size of the test <= alpha.
Jack
.
- References:
- Re: Jarque-Bera test: confidence intervals for normal data
- From: Luis A. Afonso
- Re: Jarque-Bera test: confidence intervals for normal data
- Prev by Date: Re: Jarque-Bera test: confidence intervals for normal data
- Next by Date: inverse lookup for standard normal distribution
- Previous by thread: Re: Jarque-Bera test: confidence intervals for normal data
- Next by thread: Re: Jarque-Bera test: confidence intervals for normal data
- Index(es):
Relevant Pages
|