In the case of Hill et al., they state that they "...found 11,930 genome wide significant SNPs associated with intelligence (Fig.
1a). These SNPs were found in 187 independent loci..." But the GWAS Catalog page for this study again lists a much larger number of associations: 645 in this case.
\cite{catalog}Researchers aiming to conduct further GWASs of intelligence should take these findings into account in the following ways. First, they should seriously consider not conducting such studies at all, given their poor replication record. Second, if they choose to conduct another study on this subject, they should do so on a population that is not of European descent, as such GWASs appear to be not merely lacking but entirely absent for intelligence. Finally, they should include a replication sample in their study in which they follow up on any SNPs reported to be associated with intelligence in their initial "discovery" sample.
Footnotes
1. This assumes that when a paper says "up to X individuals", where X is a number, that exactly X individuals were in fact included in the analysis. This assumption may not always be true, which is why the total sample size presented here is an upper limit.
2. The 12 lines each containing 4 SNPs are as follows:
rs35079168; rs4501664; rs11102986; rs34312136 |
rs11099040; rs13113376; rs1908039; rs1908038 |
rs13221576; rs10262915; rs6465411; rs4729127 |
rs10834449; rs12798374; rs2716458; rs1021261 |
rs4629318; rs6781149; rs17584516; rs11713158 |
rs12627933; rs11703808; rs9621305; rs761746 |
rs7475343; rs7896729; rs9423406; rs1987511 |
rs2540051; rs7589014; rs2348114; rs7563911 |
rs1329573; rs7020413; rs3824344; rs3758171 |
rs2303690; rs2560966; rs3815908; rs3936340 |
rs932541; rs6043979; rs6044003; rs6044001 |
rs528059; rs544991; rs7533254; rs16835742 |