All of the data presented above must be interpreted in light of the strikingly high variability of brain size measures across individuals (Lange et al., 1997). This high variability extends to measures of brain substructures as well. The high variability and substantial overlap of most measures for most groups being compared has profound implications for the diagnostic utility of psychiatric neuroimaging and the sensitivity/specificity in using neuroimaging to make predictions about behavior or ability in a particular individual. For example, although group average anatomic MRI differences have been reported for all major psychiatric disorders, MRI is not currently indicated for the routine diagnosis of any. Likewise, although on group average there are statistically robust differences between male and female brains, there is nothing on an individual MRI brain scan to confidently discern whether it is of a man or a woman. As an analogy, height for adult men is significantly greater than height for adult women. However, there are so many women taller than so many men that height alone would not be a very useful way to determine someone’s sex. Male/female differences in height are about twice the effect size of most neuroimaging or neuropsychological measures.
