Note that all p-values are not created equal: for a given p-value, the evidence for association increases with increasing sample size and depends on risk allele frequency. Increasing overall sample size not only increases power to detect common risk variants with modest effects, but it can increase the credibility of the observed associations. Differences in credibility by sample size for similar p-values are a consequence of the fact that these calculations take assumptions about the expected magnitude of the true allelic odds ratio into account. In particular, they assume that the true allelic odds ratio is unlikely (probability < 2.1%) to be smaller than 0.5 or bigger than 2. Since small p-values can only be achieved in small sample sizes if the estimated effect is large, these results are perceived to be less credible in this framework.
