If 1 million dollars were provided to assess the impact of marijuana decriminalization and time was of some urgency, the preferred research approach is a multistate, retrospective measurement study, preferably with a later followup to get at any sleeper effects. Secondary analyses of the extant household series data would also be pursued. If time were not a factor, the preferred research design would be a multistate long-term prospective study, probably using the combined panel and repeated cross-sectional design. Sample sizes might be kept to 500 or less per State, and presumably the sampling universe would include ages 12 and up. Secondary analyses on the national household survey would still be conducted. An advantage of the triangulation approach, particularly it it is assumed that little real effect of decriminalization is likely to be found, is that several researchers coming to the same conclusion may be more convincing than just one. On the other hand, a number of smaller and less powerful studies have a much greater risk of yielding at least one false positive conclusion. Assuming that false positives will attract far more public attention than valid negatives, then the risk of misguiding the public is substantially greater under a strategy which relies on smaller and less powerful studies. Six notes are provided.