Heterogeneity can be a major factor impacting power in, for instance, genetic analyses (McCarthy et al. 2008; Manchia et al. 2013). If the objective is to rule out potential heterogeneity prior to other analyses, then taxometrics or model-based clustering may be considered because all inter-relationships between the observed items do not have to be specified. If the results do not indicate heterogeneity, we can proceed with analyses for single homogeneous populations. Conversely, if heterogeneity is substantial, then subsequent analyses are likely to benefit from methods that take this heterogeneity into account. Detecting heterogeneity is therefore an important step in many studies, and the choice of method can be optimized by carefully considering its assumptions.
