This researcher investigated two problems: (1) the accuracy and validity of instruments which measure Piagetian levels of thought; and (2) the need for more detailed models of concrete and formal operational structures, both as they develop and consolidate, and in their most highly developed state. Several tasks such as the Piagetian correlations task are presented to demonstrate that the character or value of a task variable may create the discrepancy in performance, rather than the specific content area of a problem. It appears from these examples that it is important to analyze a task prior to presenting it. It is also indicated that the noise from instrumental errors may be larger than the noise from sampling errors. The remainder of the paper discusses the need for a more detailed conceptualization of formal thought. If it can be represented in terms of various operations and schema, it needs to be known which schema develop earliest, which depend hierarchically on the development of others, and at what level of development each must be before consolidation takes place. (HM)