Researchers, policy makers, and practitioners share a sincere interest in improving the human condition. Academics may be tempted to fault irrationality, ideology, or ignorance for the failure of research to inform policy and practice more powerfully, but policy makers and practitioners want academics to tell them "what works" in order to find a practice or method that will reliably yield desirable results. And every self-respecting researcher wants to make important contributions. Nobody is happy when the collection and analysis of empirical data on policy and practice is not respected, accessible, and put to good use. The heart of the issue lies in unrealistic expectations and aspirations, combined with a failure to take full advantage of what is possible. The research community, rightly adhering to scientific standards for causal inference, has taught policy makers to respect only the "gold standard"--findings based on experimental or quasi-experimental research, ideally employing random assignment to treatment and control groups. Action research, while it will use other methods, cannot succeed if it is less rigorous or disciplined than traditional program evaluations seeking to assess the impact of inflexibly implemented programs or interventions. It requires a clear definition of goals and a clear measurement of outcomes. It requires rigorous thinking about strategies and careful experimentation with alternative approaches when the outcomes are not satisfactory. And to become useful on a wider scale, it will require methods of communicating complex knowledge about practice and policy to others. (Contains 9 resources.)