AbstractSince 1968 more than thirty‐five social policy experiments have been conducted in the United States. During this period through 1976 these experiments were generally long‐term, large‐scale tests of major new programs; thereafter, experiments became markedly more modest in scope. Although hundreds of millions of dollars were spent on the earlier programs, the experiments probably had only a negligible impact on policies. This result stemmed from a variety of factors: Social testing actually tends to exert a conservative influence on policymaking; and the time required to complete experiments and interpret results is often incompatible with the needs of policy makers. In addition, test results are often not effectively communicated to Congress, the administration and the public, and even when the results are conveyed, policy makers are frequently skeptical about the soundness of the methodologies employed. If recent experimental programs are to exert more influence on policymaking, program oficials will need to address these dificu