Response to Irion et al. Letter
To the Editor: In their letter, Irion et al. express concern about the volumetric differences reported, stating that they are below the resolution of MRI. However, we would like to bring to their attention basic statistical principles applied in these circumstances (1), in which examination of group differences depends on the statistical properties of the sample, not on resolution in individual scans. The significance of group differences is based on the sample size, the mean group difference, and the standard deviation of the sample. Thus, even small differences can be significantly different if the measurements are repeatable (with a small standard deviation) as is the case in FreeSurfer segmentation, and if there is a relatively large sample size (N=168 in our study) and a sufficiently large difference in the means. In fact, the total volume difference measured was not small: a mean of 0.3 mm3/person×168=50.4 mm3.
1 : Two-sample hypotheses, inBiostatistical Analysis, 4th ed. Upper Saddle River, NJ, Prentice Hall, 1999Google Scholar