On Mon, 5 Aug 2002 13:43:01 +0100, Sharpe, Alan D
<[log in to unmask]> wrote:
>All,
>
>In a dosing experiment looking for differences from a control, if I am
going
>to do a multiple comparison test, e.g. Dunnett's, to ascertain which
>treatment means differ significantly from the control, What are the
benefits
>of performing ANOVA prior to this, apart from to confirm that a difference
>definitely exists?
>
>Thanks
>Alan
All
Thanks for your replies to this easy one. The responses ranged from - 'No
advantages, just one more test to provide another false positive' to 'You
must do this test first else you wont know whether to do the multiple
comparison test' with a little bit about the advantages of calculating the
pooled variance thrown in as well
As with all statistics, somewhere in there is the answer I wanted
Regards
Alan
|