Objective
This article examines the structural implications of composite variables developed in the field of “ageing”. First, using data from the Balti-more Longitudinal Study of Aging (BLSA), we illustrate how the property of feature convergence arises in practice. Second, we show how this constrains both cross-sectional stratification and longitudinal monitoring. Third, we extend the analysis to molecular implementations and demonstrate that similar structural issues persist. Fourth, we examine how these properties impact common objectives - intervention evaluation, prediction, and investigation - and show that the limitations are not confined to linear models. Finally, we consider additional properties that may help explain the widespread appeal and persistence of these models, despite clear conflicts with their stated objectives.