search

0todd0000 / spm1d

One-Dimensional Statistical Parametric Mapping in Python
GNU General Public License v3.0
63 stars 21 forks source link

Difference in results between spm1d.stats.regress and spm1d.stats.ttest_paired #25

Closed bernard-liew closed 9 years ago

bernard-liew commented 9 years ago

Dear Todd,

I am currently doing post hoc analysis using CCA and hotelling's paired ttest, using .regress and ttest_paired, respectively. I performed both using the same data, with slightly different means.

I have a vector of three components. I have three load (0,10,20) and three speed (3,4,5) combinations. For the CCA and post hoc, I wanted to answer "what is the effect of incremental load on the vector and if significant its components'. So I essentially analysed the load, independent of speed.

For the hotelling's I also wanted to answer "what is the effect of incremental load on the vector and if significant its components'. However, this time I performed one hotelling test per speed e.g. (3_0 vs 3_20), ... (5_0 vs 5_20).. If significant than I performed a paired ttest. Can I ask why this test seem to have greater power than the .regress in detecting significance for the vector's components?

Many thanks Bernard

0todd0000 commented 9 years ago

I can't see a difference between this issue and issue #24 so am closing this issue.