[Eeglablist] std_stat or statcond for mixed-designs

[Luis Fernando Ciria Perez]

Dear all,

I'm writing you with respect to the statistical analysis in EEGLAB. We are having some problems to analyze  behavioral data. We are using EEGLAB, concretly, we are using std_stat. Our design is a mixed design: 2 (Group) x 2 (Time on Task). Theoretically, we are creating a matrix as in the following example (https://sccn.ucsd.edu/svn/software/eeglab/functions/studyfunc/std_stat.m <https://sccn.ucsd.edu/svn/software/eeglab/functions/studyfunc/std_stat.m>), however the results don't look to be correct. We have analyzed these data with  classical repeated measures ANOVA and we have a clear group effect and F<1 in Time on Task effect. The results of Std_Stat are completely opposite, a clear effect of Time on Task and nothing with regard to the Group Effect. Obviously, we know results will not be exactly the same but al least, they should follow the same tendency.

Moreover, we would like to obtain p values without the mask. Do you know what could be wrong in the code? Is std_stat or statcond a good option for mixed designs?

[pcond, pgroup, pinter, statscond, statsgroup, statsinter] = std_stat(data, 'groupstats','on',...
    'condstats','on','method','perm','mcorrect','fdr','mode','eeglab','alpha',NaN,...
    'paired',{'on' 'off'},'naccu',500);


Thank you for your time.

Un saludo,

Luis Ciria


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sccn.ucsd.edu/pipermail/eeglablist/attachments/20170602/ad59026b/attachment.html>