The understanding of the political behavior of school boards can be advanced by conducting factor analysis of the voting records of trustees. While the application of these methods is generally appropriate, differences in subject matter highlight the problem of inferring political theory from numerical data. In this paper, several different political models for explaining trustee voting behavior are postulated--models which, in fact, have identical representations in terms of recorded votes and, hence, factor patterns. Further, actual voting records of two boards are analyzed in order to illustrate two different types of voting behavior. Explanations for these differences require additional knowledge concerning variables other than the votes themselves. Finally, there is a discussion of some of the modes of analysis currently in use, with recommendations as to the best alternatives. (Author/DN)