Recent federal education policies and reports have generated considerable debate about the meaning, methods, and goals of "scientific research" in mathematics education. Concentrating on the critical problem of determining which educational programs and practices reliably improve students' mathematics achievement, these policies and reports focus almost exclusively on experimental/quasi-experimental research. Partly because of this focus, but partly because of adherence to a particular underlying philosophy, these policies and reports advocate an extremely narrow view of "scientific" research. This view excludes many high-quality research methods and a majority of extant mathematics education research studies that can and should be used in efforts to guide and improve mathematics education. In this article, the authors argue that there are numerous rigorous scientific research methods beyond experiments/quasi-experiments, and they illustrate how different types of research are needed for different constituencies and purposes in mathematics education. (Contains 4 figures, 1 table and 9 footnotes.)