This article deals with alternative research strategies for developmental psychopathology. It argues that most applications of statistical methods in empirical research are variable centered, not person oriented. As a result, conclusions are often drawn that fail to do justice to the diverse nature of populations. It is recommended that we take seriously the importance of the implications of data aggregation. The difficulties of making inferences from a more aggregated level of analysis to a less aggregated level are explained and exemplified. We explain that a set of variables displays dimensional identity if the variable relationships remain unchanged across the levels or categories of other variables. Data examples of intelligence divergence and of Child Behavior Checklist subpopulation differences show that lack of dimensional identity can lead to incorrect conclusions. Schmitz' theorems on aggregation and the validity of results at the aggregate level for individuals are illustrated using data from a study on the development of alcoholism and discussed from a person-oriented perspective. Statistical methods suitable for person-oriented data analysis are reviewed.