| |
 |
|
|
Science Forum Index » Statistics - Math Forum » Paired t test with multiple comparisons
Page 1 of 1
|
| Author |
Message |
| Francogrex |
 Posted: Sat Apr 05, 2008 6:49 am |
|
|
|
Guest
|
If I'd like to do a paired t test but adjust for multiple comparisons
using something like the method of bonferroni, is it enough to
multiply the alpha of each two-by-two paired t test by the total
number of comparisons (say I have 3 comparisons so if in one test the
alpha is 0.05 then I would correct it as 3*0.05=0.15), just like for
unpaired t test with multiple comparisons? Or should the approach be
different? Thanks |
|
|
| Back to top |
|
| Ryan |
| Posted: Sat Apr 05, 2008 3:15 pm |
|
|
|
Guest
|
On Apr 5, 12:49 pm, Francogrex <fra...@grex.org> wrote:
Quote: If I'd like to do a paired t test but adjust for multiple comparisons
using something like the method of bonferroni, is it enough to
multiply the alpha of each two-by-two paired t test by the total
number of comparisons (say I have 3 comparisons so if in one test the
alpha is 0.05 then I would correct it as 3*0.05=0.15), just like for
unpaired t test with multiple comparisons? Or should the approach be
different? Thanks
In order to make the Bonferroni adjustment, you would either (1)
divide alpha by the number of comparisons or (2) multiply the p-values
by the number of comparisons.
Ryan |
|
|
| Back to top |
|
| |
|
Page 1 of 1
All times are GMT - 5 Hours
The time now is Thu Jul 24, 2008 5:00 pm
|
|
