During the past several years interest in using clinical interviews to assess students conceptual knowledge has increased. However, using clinical interviews when research questions require quantitative comparisons has been difficult. The central difficulty is to construct variables that can be used to quantify and statistically compare the results of interviews while maintaining exactly what student conceptions are associated with each variable. The purpose of this article is to present a technique for constructing variables that overcomes this difficulty. The technique involves the representation of each proposition a student uses in terms of a standard set of underlying predicates, concepts, and relationships among concepts. Each proposition represented in these standard terms can be used as a variable to quantitatively compare the knowledge of students as expressed in clinical interviews.