CRITICS OFTEN HOLD THAT RANDOMIZED FIELD EXPERIMENTS ARE IMPOSSIBLE TO IMPLEMENT IN THE 'REAL WORLD'; THAT THEY ARE EXPENSIVE AND SLOW; THAT THEY CAN BE REPLACED BY STATISTICAL ADJUSTMENT OF NONEXPERIMENTAL DATA; AND THAT THEY ARE UNETHICAL. THIS ARTICLE EXAMINES FOUR BROAD CLASSES OF CRITICISM IN LIGHT OF THE AUTHOR'S EXPERIENCE IN ACCUMULATING VALID INFORMATION ABOUT PROGRAM EFFECTS. THE CRITICISMS BEAR ON FEASIBILITY, SCOPE, USEFULNESS, AND ETHICALITY OF RANDOMIZED EXPERIMENTS FOR EVALUATING SOCIAL PROGRAMS. THROUGH THIS REBUTTAL THE AUTHOR OFFERS SUGGESTIONS FOR IMPROVEMENT OF RANDOMIZED FIELD EXPERIMENTS.