MEASURED VARIABLES, TREATED AS SURROGATES FOR UNMEASURED CONSTRUCTS OF THEORETICAL INTEREST, OFTEN ARE CONTAMINATED BY SOURCES OF VARIATION OTHER THAN THOSE THEY ARE INTENDED TO REFLECT. IN STRUCTURAL EQUATION MODELS, SUCH INDICATOR CONTAMINATION MAY SERIOUSLY DISTORT PARAMETER ESTIMATES AND AFFECT THE VALIDITY OF SUBSTANTIVE CONCLUSIONS. AN ANALYSIS OF INDICATOR FLAWS IS RELEVANT TO THE ACHIEVEMENT OF MORE ACCURATE PARAMETER ESTIMATES AND TO THE IMPROVEMENT OF INDICATORS AND INCREASED CLARITY IN THEORETICAL CONCEPTS. A MODEL OF AND DATA PERTAINING TO THE INDUSTRIAL AND POLITICAL DEVELOPMENT OF NATIONS ARE USED TO EVALUATE THREE INDUSTRIAL DEVELOPMENT INDICATORS (GROSS NATIONAL PRODUCT, ENERGY, AND LABOR DIVERSITY) AND SIX POLITICAL DEVELOPMENT INDICATORS (EXECUTIVE FUNCTIONING, LEGISLATIVE FUNCTIONING, PARTY ORGANIZATION, POWER DIVERSIFICATION, CITIZEN INFLUENCE, AND AN INDEX OF POLITICAL REPRESENTATION). CORRELATION DATA FROM THESE INDICATORS ARE PRESENTED, AND INTUITIVE INFERENCES FROM THE PATTERN OF RESIDUALS IN CORRELATIONAL ANALYSIS ARE DISCUSSED. TWO-INDICATOR AND THREE-INDICATOR SUBMODELS ARE CONSTRUCTED TO MINIMIZE INDICATOR FLAWS. THE THREE-INDICATOR SUBMODEL PROCEDURE IS APPLIED TO ACTUAL DATA. THE RELIANCE OF THE PROPOSED STRATEGY ON THE EXAMINATION OF FIT AND RESIDUALS IS EXAMINED. SUPPORTING DATA AND EQUATIONS ARE INCLUDED. (DEP)