This paper offers concrete examples from recent policy research about how qualitative and quantitative methods can be combined to better address research questions. Using a conceptual framework developed by Rossman and Wilson (1985), later expanded by Greene, Caracelli, and Graham (1989), the paper discusses how the design and analysis phases of research can apply mixed methods to enhance understandings of social phenomena. A purposive sample of 57 mixed-methods evaluation studies was used to test components of the framework. A new typology of mixed-methods purposes is developed that identifies: (1) corroboration; (2) elaboration; (3) development; and (4) initiation. Crossing the four purposes with research design or analysis stages yields a 4 x 2 matrix. Examples of how methods can be effectively combined are drawn from two policy studies that focused on the effects of state policies on local school districts. In both studies a combination of qualitative and quantitative methods was consciously included. Mixed-methods approaches work best when more than one researcher is engaged, when norms of respect and collegiality prevail, and when an attitude of healthy skepticism about both theory and method exists. (Contains 25 references.) (SLD)