I think your question is, if you used a power calculation based on test A to
decide what sample size you should use, is it permissible to use test B to
decide whether your result is significant?
Obviously the question is only relevant if A & B give different results. In
some cases this ought not to happen because the two tests are closely
related and depend on the same assumptions.
In other cases A and B could differ in their sensitivity. This usually
means that the tests require different assumptions about the data. You must
have a very good reason for changing assumptions after you have designed an
experiment.
Hope this helps
James Rothman
James Rothman Economic & Marketing Research
[log in to unmask]
----- Original Message -----
From: "Shankar Chivukula" <[log in to unmask] >
To: <[log in to unmask]>
Sent: Friday, July 06, 2001 9:07 AM
> Dear Assume list members,
>
> Is there is anyinformation or any book which says that
> whatever the statsitical tool we use to calculate the
> sample size, should the same tool be used for analysis
> also.
>
> hope i am clear
>
> or
>
> suppose if i calculate a sample size using anova can't
> I analayse the data using a t-test?
>
> Please help me in clarifying my doubt.
>
> Thanks
> Srikanth
>
> __________________________________________________
> Do You Yahoo!?
> Get personalized email addresses from Yahoo! Mail
> http://personal.mail.yahoo.com/
>
|