Hi Anderson,
Thanks for the explanation. Would the following be accurate to do the following:
1) I first use the -twotail option as a first level F test to see which of my comparisons show a statistically significant difference.
2) For those contrasts that show something significant on the -twotail test above, I run the -corrcon with the one tail tests. This way I am not losing statistical power by correcting for contrasts that do not show anything significant on the first level.
Thanks,
Ajay
|