In addition, it discusses issues of interpretation of genetic association studies, with a particular focus on dealing with genotyping error, correction for multiple comparisons, population stratification, and power.
Statistical inference was performed at the voxel level, with both a family-wise error correction for multiple comparisons (p < 0.05) and an uncorrected threshold (p < 0.001; cluster size:100).
For example, if the error rate is 30/10,000 in one site, and the error rate in the second site is triple (difference = 60), then at 80% power (and not adjusting the overall type I error rate for multiple comparisons), we would require 2900 fields per group.
Results did not survive family-wise error correction for multiple comparisons, and hence were assessed uncorrected at P < 0.001.
Statistical parametric maps were thresholded at the cluster level using family-wise error correction for multiple comparisons at P < 0.05.
A voxel-wise statistical threshold of P < 0.05, family-wise error corrected for multiple comparisons was applied in all analyses.
Analyses of endpoints and treatment comparisons were performed in a fixed sequential order to control the type 1 error rate for multiple comparisons.
In 31/33 analyses, the effect sizes for PIQ were significant (p < 0.05 in extent) after family wise error correction for multiple comparisons across the whole brain.
The regions stimulated in that study were the only ones for which the statistical significance of the reported correlations survived a family-wise error correction for multiple comparisons.
The voxel-level statistical threshold was set at P < 0.05, with familywise error correction for multiple comparisons and a cluster extent threshold of 10 voxels.
In all statistical tests, the results were considered statistically significant if the P value was less than 0.05 after family-wise error correction for multiple comparisons.
