mne.stats.f_mway_rm

mne.stats.f_mway_rm(data, factor_levels, effects=’all’, alpha=0.05, correction=False, return_pvals=True)[source]

Compute M-way repeated measures ANOVA for fully balanced designs.

Parameters:

data : ndarray

3D array where the first two dimensions are compliant with a subjects X conditions scheme where the first factor repeats slowest:

            A1B1 A1B2 A2B1 A2B2
subject 1   1.34 2.53 0.97 1.74
subject ... .... .... .... ....
subject k   2.45 7.90 3.09 4.76

The last dimensions is thought to carry the observations for mass univariate analysis.

factor_levels : list-like

The number of levels per factor.

effects : str | list

A string denoting the effect to be returned. The following mapping is currently supported (example with 2 factors):

  • 'A': main effect of A
  • 'B': main effect of B
  • 'A:B': interaction effect
  • 'A+B': both main effects
  • 'A*B': all three effects
  • 'all': all effects (equals ‘A*B’ in a 2 way design)

If list, effect names are used: ['A', 'B', 'A:B'].

alpha : float

The significance threshold.

correction : bool

The correction method to be employed if one factor has more than two levels. If True, sphericity correction using the Greenhouse-Geisser method will be applied.

return_pvals : bool

If True, return p values corresponding to f values.

Returns:

f_vals : ndarray

An array of f values with length corresponding to the number of effects estimated. The shape depends on the number of effects estimated.

p_vals : ndarray

If not requested via return_pvals, defaults to an empty array.

Notes

New in version 0.10.