Import
LIMO_EEG analyses data for all electrodes and data points – because it is
computationally demanding we also provide the option to only analyze a portion
of the data. Typically, the end of your spectrum is not real due to filtering
and this makes sense to restrict the data to this range. Similarly, in the time
domain you can restrict the analysis range.
Specify
This is where you specify which conditions were
presented to the subjects.
A categorical variable is a variable
which has various levels as obtained e.g. in factorial designs. At this stage
LIMO_EEG requires a text or mat file (usually coming from the log file of your
experiment) describing those levels. In a typical 2*2 experiment with levels A1
A2 B1 B2, you would produce a text or mat file describing which condition was
presented at each trial like e.g. 1 1 1 2 2 3 3 4 4 3 2 1 2 3 4 3 2 1 1. LIMO
will interpret this as 4 conditions and re-organize trials accordingly. Most of
the time, the best way to describe 1st level analyses is to describe each
condition separately. However, you can specify factors rather than condition.
In that case you .txt file or .mat file needs to contain 2 columns like e.g. 1
1 1 2 2 1 2 2 2 1 1 2 in the 1st column for the 1st factor with 2 levels and
e.g. 1 2 2 1 2 1 2 1 2 for the 2nd factor with 2 levels. Don't worry about the
actual numbers you use, LIMO only use then to index conditions so it doesn't
have to be 1 2 3 etc ..
Full factorial: if more than 1 factor is entered (i.e. the
categorical file used has more than 1 column) this option
become available, allowing to create an interaction term.
A continuous variable is a variable
which varies across all trials like in regression or parametric designs. At
this stage LIMO_EEG requires a text or mat file (usually coming from the log
file of your experiment) describing these variations. This file can also contain
multiple columns, for multiple
variables.
There is also a box 'do not z-score
regressors' which becomes available once the continuous regressor
file is loaded. Continuous variables are always standardized as to reflect
effects expressed as standard deviations per voltage. In addition, scaling the
variables allow meaningful comparisons between them. It is recommended to
always leave this box unticked unless you have good reasons (like having zscored your variable already – to split a variable per
condition, see limo_tools).
Specify
For now the type of analysis availbale is only 'Mass-univariate' using 'OLS' (ordinary
least squres). In future release multivariate
analyses and weighted analyses will be available. These boxes are deactivated.
Activate at your own risk.
Bootstrap data and tfce
are available for single subject. For stamdard
hierarchical modelling, only Beta parameters (or combinations of them) are necessay, so there is no need to estimate the null
distribution per subjects. However, if you want to test effects within subject,
you need to correct for multiple comparisons and bootstrapping is necessary.
Done
This will execute commands to create some files
- in particular a LIMO.mat structure containing
information about the subject, data, etc and Yr which the same data that are contained in the .set but
now reorganized according to your design matrix - following pressing done the
design matrix pop up and it you are happy with the design press 'yes' to get
the analysis going