In addition to the limitations of direct empirical evidence, certain caveats regarding subtyping research in general should be noted. The analytical methods used to identify subtypes (e.g., cluster analysis) are primarily descriptive in nature and intended to uncover typological structure in empirical data. Although the solutions that these methods produce can aid in the development of subtyping theory and inform the design of prevention and treatment programs, the results are primarily of heuristic (i.e., educational) value. The subtypes suggested by such methods reflect only the characteristics of the sample studied and the investigator’s choice of relevant personal attributes. Thus, researchers should avoid the tendency to construe subtypes as more than a theoretical construction. Specific subtyping methods are particularly well suited to certain tasks; different approaches (e.g., two versus four clusters for the same study population) are not necessarily competing representations of “reality.” The usefulness of any particular solution will depend on the investigator’s purpose (for further discussion of this general point, see Del Boca 1994).
