Our results highlight the value of representative cohorts (including birth cohorts) where there is little or no selection into the cohort. The issue of whether the study is intended to be representative in the first place is a somewhat separate issue, albeit a related one (see Box 3 Does representativeness matter?Representativeness refers to the comparison between the population about which we wish to draw inferences (e.g. all people in the UK, all men with a prostate cancer diagnosis), and what we have described as the “intended study population” (e.g., doctors in the British Doctors Study, all men diagnosed with prostate cancer, etc.). The degree to which the intended study population can be used to draw inference about the population of interest will depend on factors influencing selection into the intended study population. For example, the biology of doctors in the British Doctors Study will be similar to the biology of all people, so this intended study population should enable us to draw inferences about many exposures and outcomes that would hold for the general population of the UK (and indeed
