Hello! We just uploaded to biorxiv.org
our manuscript with an analysis of the corpus callosum (CC) and brain volume (BV) in Abide (N~700!!):http://biorxiv.org/content/early/2014/02/15/002691
First, we did a meta-analysis of 17 papers, 12 of which reported a smaller CC in autism. However, the average statistical power was very low: 20% on average. So, we looked at the CC in Abide. We had 99% power, but we did not find any difference!!. The previous findings could be due to several reasons: publication bias, researcher's degrees of freedom, etc. But there maybe also something in the strategies researchers use to analyse their data that introduced artefactual differences. We looked in particular to the relationship between CC and BV, and observed that scaling was not linear: larger brains tend to have proportionally smaller CCs. We also observed the quite well known correlation between IQ and BV, but only in the control group. Very often researchers normalise CC, and use CC/BV instead of BV, as a way of eliminating the effect of BV. However, because of the allometric scaling of CC, this will not eliminate a BV effect. Another common strategy is to match autistic and control group by IQ. But because the correlation is not the same in both groups, this may bias the recruitement to controls with smalle BV and patients with large BV. We tested the conditions under which these effects could induce detectable differences in CC size using simulations. We are looking forward to your comments!