# [Mne_analysis] Group level ERF regression

Jona Sassenhagen jona.sassenhagen at gmail.com
Tue Sep 3 15:30:31 EDT 2019
 Previous message: [Mne_analysis] Group level ERF regression Next message: [Mne_analysis] MEG North America and MEG Hackathon - Abstracts Due Sep 13 Messages sorted by: [ date ] [ thread ] [ subject ] [ author ] Search archives:

```        External Email - Use Caution

For most purposes, you can treat the coefficients from a linear model just like normal ERPs (Which would correspond to coefficients from an intercept-only model run on the subset of epochs). For an explanation, see eg Smith & Kutas 201x psychophysiology.

________________________________
From: Alexandre Gramfort <alexandre.gramfort at polytechnique.org>
Sent: Tuesday, September 3, 2019 9:09 PM
To: Denis A. Engemann; Jona Sassenhagen
Subject: Fwd: [Mne_analysis] Group level ERF regression

for you ! or José...

A

---------- Forwarded message ---------
From: Kambiz Tavabi <ktavabi at uw.edu<mailto:ktavabi at uw.edu>>
Date: Tue, Sep 3, 2019 at 7:50 PM
Subject: [Mne_analysis] Group level ERF regression
To: mne_analysis at nmr.mgh.harvard.edu<mailto:mne_analysis at nmr.mgh.harvard.edu> <mne_analysis at nmr.mgh.harvard.edu<mailto:mne_analysis at nmr.mgh.harvard.edu>>

External Email - Use Caution

Hi all,
I'd like to carry out single-trial regression using linear_regression python function to evaluate effect of continuous variable on ERFs. The following gist yields the desired output for a single subject:

n_obs = len(epochs)
intercept = np.ones((n_obs, 1))
# Design matrix for continuous covariate
regressor = np.linspace(8, 66, n_obs)
dmat = np.concatenate((regressor[:, np.newaxis], intercept), axis=1)
# regression between ERF & covariate
res = linear_regression(epochs, design_matrix=dmat, names=names)

But it's not clear to me what to do with the resulting mne.Evoked objects containing beta values, to describe the regression results across a group of subjects. E.g. does it make sense to average the evoked (beta) objects across the group? and if so how do I derive the t-values? Thanks in advance.

Kambiz Tavabi PhD
ktavabi at uw.edu<mailto:ktavabi at uw.edu>
206-685-6173<tel:206-685-6173>
Institute for Learning & Brain Sciences
1715 Columbia Road N Portage Bay Building
Box 357988
University of Washington