Checking 1st level weights

1st level weights

The default behaviour of LIMO is to use Weighted Least Squares, thereby each trial is assigned a weight reflecting how much it's dynamic differs from the other trials. It is useful, and some insight can be gained as well, to check weights and trials of low and high weights.

LIMO tools --> check single trial weights --> select a list (.txt) of LIMO files and your channel location file

Subjects weighted data

A structure (.mat) is computed and can be examined using the Plot central tendency and differences - the dimensions are such that all subjects data are displayed for a given weight decile.

Bias

A ANOVA/t-test is computed on weights across all conditions, reveling if some conditions have different weight from the others, which should not be the case - unless the statistical model is mis-specified.

Trial differences

A t-test is computed comparing low vs. high trials across all conditions - thus ensuring there is no spatial bias as there is no a-priori reason for low weight trials to appear more often at a given location.

Command line

limo_CheckWeight(list_of_LIMO.mat,expected_chanlocs,'CheckBias','on','TestDifference','on','SingleSubjectAnalysis','on','PlotRank','on')

There is on option only available through command line: 'SingleSubjectAnalysis' which allows to compute differences between low and high weight trials, subject-wise. This adds to the trial differences analysis which expect some consistency between subjects (as do any group level stats).