Hi,
As part of SPSS Repeated Measures it performs what it calls
pairwise comparisons of the main effects which, in my case
at least since the repeated measure is the only thing in the model,
is the same as performing multiple paired t-tests.
There is an option for a bonferroni correction and I have compared the
results I obtain with this on and off.
There are 6 tests and so with the Bonferroni correction the p-values are
all
multiplied by 6 - I'm OK with this.
When it comes to the Confidence Intervals with the correction both on and
off
it claims to report the 95% intervals, however after a few calculations
it seems that with the correction on the intervals are actually 99.166666%.
I am happy with the figure of 99.166666% but don't see why SPSS should
claim
these are 95% intervals.
My question is, if I performed 6 paired t-tests (4 scores all compared in
pairs)
should I report:
1) 99.166666% CIs with the standard p-values
2) 99.166666% CIs with the p-values all multipled by 6
3) 95% CIs with the p-values all multipled by 6
and what should I say I am reporting, e.g. '95% CIs with Bonferroni
correction'
Many thanks
Jon
|