population-based samples can provide more power. If the case definition is a relatively rare sub-phenotype that shows clear segregation in families (and families with multiple affecteds can be collected and genotyped), then a family-based approach will be preferable. A second condition before embarking on a population-based approach is the possibility that one or more of the underlying genetic variants could be common (e.g., with a POPULATION ALLELE FREQUENCY >0.05). Moderately rare (frequency 0.01-0.05) variants can also be detected with available sample sizes but only if they carry a large effect (relative risks > 2.0). If there is an a-priori hypothesis that all undetected genetic variants are rare and of small effect, the samples sizes required to detect such effects in a population-based study will be unfeasibly large 5.
