Ford ’ S Procedure for Combining Multiple Sets of Partially Ordered Data 1 Frank M . Andrews

MORE than a decade ago, Lester R. Ford, Jr. published a mathematical procedure for combining multiple sets of partially ordered data (Ford, 1957). Although this procedure has received little attention from social scientists, it solves a methodological problem which is not uncommon in some fields of educational and psychological research. Furthermore, it permits certain flexibilities in the way data are collected which, if used, would probably increase measurement validity in some kinds of variables. Ford’s procedure is likely to be useful whenever one wants to achieve a single rank order by combining data obtained from multiple &dquo;judges&dquo; who have placed &dquo;objects&dquo; into ordered categories. Of course, the &dquo;judges&dquo; need not be people, but any procedure which orders objects. The procedure has special advantages when data are untidy in any of several ways: (a) some objects are unclassified by some judges (e.g., when some judges lack knowledge about some of the objects being ranked), (b) there is less than perfect agreement among the judges, (c) some objects