Adjustments for multiple tests, like Bonferroni correction, are needed for medical association studies but may create more problems.
Unlike a multiple testing procedure like PACT, which accounts for multiple SNP tests in order to properly control the type I error, ARTP combines information across SNPs within a gene or a pathway in order to increase the power to detect a gene or pathway level effect.
We performed Bonferroni-like multiple testing correction [ 33] to determine significance thresholds for GWAS.
One SNP (rs11894053) was nominally associated (P < .05, uncorrected for multiple testing) with social autistic-like traits.
We performed Bonferroni-like multiple testing correction (Matthies et al. 2014) to determine significance thresholds for GWAS.
To globally correct enrichment P-values to control family-wide false discovery rate (e.g., <= 0.05), DAVID provides multiple testing correction techniques like Bonferroni, Benjamini and FDR, which were taken into account during the analysis.
One should keep in mind that correction for multiple testing in situations like the present one, with several (related) end points and correlations between exposures, is complex (Bender and Lange 2001) and that the FDR approach that we have used assumes independence between covariates.
Applying strict corrections for multiple testing would make investigations like the present impossible to perform, because a sample big enough to withstand such a correction procedure would probably never be possible to obtain.
As no Bonferroni-like adjustment for multiple testing was applied during this exploratory analysis, P-value nearing 0.05 has, however, to be considered cautiously.
Despite these results, when the Hochberg method for multiple testing is not applied, as authors like K. Rothman suggest[ 50], not only undesirable LEs are statistically significant, also family events.
