I agree that rho=0 as typically used is silly. Well are
you are arguing then for the Bayesian framework of getting a probability
distribution on rho.??
dennis roberts wrote:
>
> At 11:58 AM 4/11/00 -0400, you wrote:
> >at the bottom
> >
> >
> >it would be easy to show whether rho is more likely closer to 0 or .5
> >given the parameters of a designed study
>
> granted ... but all this says is that the traditional null of 0 ... and a p
> value associated with IT ... is rather useless, would you not agree? or,
> perhaps ... it IS useless IF you are interested in what rho is?
>
> >what if the confidence interval for rho was (0,0.5), is rho closer to 0
> >or .5
> >or .25 for that matter?
>
> right .. what if the ci were ... .4 to .7? or .3 to .6? or .4 to .8?
>
> do we need a planned study to help answer this?
> now, i agree that a planned study ... puts us in a better position of
> getting information we want ... but the failure to do so does not NOT let
> us get that information ... where the classic null of rho = 0 in this case
> ... if rejected ... does not provide us with information we want
>
> which gives us information about what rho possibly is ...
>
> A. reject rho = 0 at .01 ... or,
>
> B. ci for rho is ... .38 to .65?
>
> i say B tells us something about what rho might be ... and A does not
A does tell us where rho might because A also involves knowledge
of location and scale. When we test A we certainly know
the estimate of rho. Then the test involves a distance of
the estimate of rho from the hypothesized rho as measured
by the pvalue. The bigger the estimate of rho is from the
hypothesized rho the smaller the pvalue.
A and B are not independent quantities if we consider p-values in the
mix
> the question here is not is it 0 ... but what it is ... now, for sure .. it
> will NOT tell us what rho IS ... but, we have AN idea, rejecting the null
> of 0 helps us not a bit other perhaps AT most to say we don't think it is 0
===========================================================================
This list is open to everyone. Occasionally, less thoughtful
people send inappropriate messages. Please DO NOT COMPLAIN TO
THE POSTMASTER about these messages because the postmaster has no
way of controlling them, and excessive complaints will result in
termination of the list.
For information about this list, including information about the
problem of inappropriate messages and information about how to
unsubscribe, please see the web page at
http://jse.stat.ncsu.edu/
===========================================================================
