Different ways of creating dependent variables for statistical analysis in...
Brought to you by:
jdien
Menu
▾
▴
Hi there,
Usually when using the toolkit, I use the windowing feature to derive mean amplitudes for each observation for each factor. One thing I noticed is that, for a one step temporal PCA, it seems like the choice of time window for windowing doesn't have any effect on the F stat and pvalue of the ANOVA that I run on these mean amplitudes. I was assuming this was because the amplitude is really just the factor loading x the factor score for a given observation such that the factor score for each observation is just being scaled by the same value (the portion of the factor loading defined in the windowing process). However, when I directly extract the factor scores, and compare the ANOVA for these raw factor scores with the windowed mean amplitudes, I get very different ANOVA results. Did I just make a mistake somewhere in my extraction/analysis of the factor scores or is it true that the factor scores will lead to different results as compared to the windowed amplitudes?
Best,
Victor
Yes, that is exactly right re windowing. You should indeed get the same results. How did you go about getting the factor scores? And what version of EP Toolkit are you using? More information required to answer question.
Thanks for the response! I am using version 2.91 of the toolkit. I got the factor scores by pressing the "edit" button and then navigating to the FacScr panel in the "PCA" tab. I used the export button to save a .txt file and then made a script that adds electrode, condition, and subject information in separate columns. It's certainly possible that I made a mistake during this step, but I just looked over the code and spot checked a few values and everything seems to be working correctly... I'm going to keep checking for mistakes I could've made, but really I just wanted to check that I was correct to be concerned about my factor score results not lining up with the windowed mean amplitude results.
Thanks again!
Victor
The problem is that you need to update to the current version (2.93). It fixes a bug in that function. Always make sure to be using the current version.
Joe
Ok that fixed my problem thanks so much!