IT IS NOTED THAT ORDINAL DATA IS THE TYPE OF RELIABLE INFORMATION MOST FREQUENTLY AVAILABLE TO SOCIAL SCIENTISTS. ORDINAL DATA PROVIDE AN ORDERING OF CASES FROM THE HIGHEST TO LOWEST, HOTTEST TO COLDEST, AND BEST TO WORST. ALTHOUGH ORDINALS LACK A UNIT OF MEASUREMENT, THEY PERMIT A COMPARISON OF ITEMS. THE BASIC STRATEGY OF THE PAPER IS TO DEVELOP A GENERAL PREDICTION FRAMEWORK AND A RELATED STATISTICAL MEASURE IN THE HANDLING OF ORDINAL DATA. THIS GENERAL FRAMEWORK IS THEN USED TO EXPLAIN THE MOST WIDELY USED MEASURES OF BIVARIATE ASSOCIATION FOR ORDINAL VARIABLES, INCLUDING GAMMA AND THE VARIOUS DELTA AND TAU MEASURES. IT IS NOTED THAT MANY OTHER TYPES OF PROPOSITONS CAN ALSO BE EVALUATED WITH THE GENERAL FRAMEWORK. AFTER DEALING EXTENSIVELY WITH BIVARIATE PROPOSITIONS, THE EVALUATION OF MULTIVARIATE PREDICTIONS IS BRIEFLY CONSIDERED. THE GENERAL FRAMEWORK IS CONSIDERED TO PROVIDE NOT ONLY A COMMON PERSPECTIVE FOR INTERPRETING STANDARD MEASURES OF ASSOCIATION, BUT ALSO TO PERMIT CUSTOM DESIGNING OF A MEASURE WHEN A RESEARCH APPLICATION ARISES WHERE NO STANDARD MEASURE IS APPROPRIATE TO EVALUATE THE SPECIFIC PROPOSITION BEING CONSIDERED. (AUTHOR ABSTRACT MODIFIED)