SOME EXPERIMENTAL PROGRAMS ARE IMPOSSIBLE TO ESTABLISH IN A GIVEN CULTURAL MILIEU BECAUSE OF THE POTENTIAL OF POPULAR OUTCRY. PRIVATE OR COOPERATIVE PROGRAMS OUTSIDE THE GOVERNMENT CAN SOMETIMES CONDUCT EXPERIMENTAL PROJECTS, HOWEVER. PRACTICAL DIFFICULTIES INVOLVED IN SOCIAL INTERVENTION EFFORTS SUCH AS ADMINISTRATIVE INEFFICIENCY, AMBIGUOUS TRANSMISSION OF INSTRUCTIONS THROUGH THE ADMINISTRATIVE HIERARCHY, INADEQUATE RESOURCES, INSUFFICIENT POPULAR SUPPORT, AND INAPPROPRIATE CHOICE OF POLICY CAN HAMPER THESE EFFORTS. THESE DIFFICULTIES ARE FURTHER COMPOUNDED BY MOST AGENCIES' FAILURE TO COLLECT EVALUATION DATA OR THEIR COLLECTION OF POLITICALLY INFLUENCED DATA. OFFICIAL CRIME STATISTICS, FOR EXAMPLE, ARE INFLUENCED BY STATES' AND CITIES' POLITICAL DESIRES TO PROMOTE ECONOMIC GROWTH AND TOURISM. EVALUATION IS FURTHER HAMPERED IN SITUATIONS IN WHICH A CONTROL GROUP IS INAPPROPRIATE OR UNETHICAL. THE UNTREATED CONTROL GROUPS IN A U.S. PUBLIC HEALTH SERVICE SYPHILLIS TREATMENT EVALUATION IS ONE OF THE MOST SERIOUS ETHICAL CASES YET DISCOVERED. AFTER AN EVALUATION IS COMPLETE, IT MAY BE IGNORED FOR PRACTICAL OR POLITICAL REASONS. TIME LAG INVOLVED, POPULAR DEMAND, PREMATURE PROGRAM EVALUATIONS, OR SUPPRESSION OF RESULTS MAY LIMIT AN EVALUATION'S USE. IN ADDITION, TRADITIONAL BUREAUCRATS ARE RELUCTANT TO BE EVALUATED. EVALUATION IS NOT A SIMPLE PROCESS. ITS PROBLEMS HAVE LED SOME PEOPLE TO CONCLUDE THAT EVALUATION RESEARCH RESULTS ARE MEANINGLESS AND SUCH RESEARCH SHOULD THEREFORE CEASE. NEVERTHELESS, FEW EVALUATION PROJECTS ENCOUNTER MORE THAN A SMALL FRACTION OF THESE POLITICAL DIFFICULTIES. MEANS TO OVERCOME TECHNICAL DIFFICULTIES CONTINUE TO BE FOUND. TO AVOID RELYING ON ARGUMENTATIVE POWER, COMMONSENSE LOGIC, OR ANECDOTAL EXAMPLES, AGENCIES SHOULD CONTINUE EVALUATION RESEARCH. FOOTNOTES WHICH INCLUDE REFERENCES ARE INCLUDED. (CFW)