next up previous
Next: Test for Gaussianity of Up: Results and Discussion Previous: Intersession Effect Size Plots

Quantification of Intersession Variance

In order to better quantitate the size of intersession variance, we estimated the mean ratio of RE (ME minus FE) to FE variance. Note here that any comparison between the RE and FE variance will be dependent on the number of time points in each session, with a larger number of time points leading to an increase in the RE:FE ratio.

The results are shown in Table 2. The interpretation of this is simple yet important; in these datasets, intersession variability is not large compared with within-session variability.6

Note that we cannot make very useful interpretations of the variations across methods of the variance ratio, particularly without also taking into account the estimated effect size; hence the use of mixed-effects Z for the main method comparison results shown below in Section 4.6.


Table 2: Mean estimated ratio of RE (intersession) variance to FE (within session pooled) variance.

  A C D E F G
cognitive 1.0 0.3 1.5 0.5 0.6 1.0
motor 0.9 0.3 1.4 0.6 0.6 1.1
visual 1.4 0.3 1.9 0.8 0.8 1.7



next up previous
Next: Test for Gaussianity of Up: Results and Discussion Previous: Intersession Effect Size Plots