Assessment within the fields of early childhood education and early childhood intervention is guided by the deductive-psychometric model, which is a framework for legitimizing constructs that arise from theories. An alternative approach, termed the inductive-experimental model, places significantly more restrictions on what constitutes a legitimate construct. In this paper, the utility of these two assessment models, one more generative and one more restrictive, are evaluated within the context of a Head Start setting. Given the pragmatic goal of informing instruction, we argue for the superiority of the more restrictive approach. Implications for early childhood intervention are also discussed.