Selection bias is pervasive in human studies and even more so for epigenome-wide association studies (EWAS) due to the relatively high costs of measuring DNA methylation. Although this bias is often acknowledged, its impact on findings has not been evaluated. We aim to perform such an evaluation for EWAS performed in ALSPAC. We will characterise selection by comparing the characteristics of ALSPAC participants measured at baseline with and without DNA methylation profiles and then estimate the resulting bias using simulations informed by the data.