On the robustness and validity of groups

In response to Nagin's comment on our paper, Methodological Sensitivities to Latent Class Analysis of Long-Term Criminal Trajectories,'' we reconsider the robustness and validity of group-based approaches to criminal trajectories and introduce additional issues for future research. We emphasize the limitations of typological approaches and the dangers of reifying the idea of distinct offender groupings for research and policy. We also clarify misunderstandings about the use of hierarchical linear models for studying trajectories of crime. Our basic conclusion is that methods in criminology need to be more tightly linked with theory.