Scienti¢c management of science?

A framework is advanced for allocating research resources, based on the value of the information that proposed projects are expected to produce. As an organizing device, the framework uses integrated assessments, showing the relationships between the predictors of outcomes arising in important decisions (e.g., interest rates, mortality rates, crop yields, crime probabilities). Proposed projects are evaluated in terms of their ability to sharpen estimates either of those variables or of the relationships among them. This approach is intended to allow diverse forms of science to show their potential impacts ^ while encouraging them to integrate their work. Where suitable estimates are available, the expected value of the information from alternative studies can be computed and compared. However, even at a qualitative level, formal analyses can improve the e¤ciency and integration of research programs. Scienti¢c management of science? Science requires resources. At the very least, it needs the time and energies of skilled individuals. Often, it requires money as well ^ in order to secure the facilities, release time, travel expenses, and other things that money can bring. As a result, allocating resources is a task that, in one way or another, occupies all scientists. As individuals, they must make choices about which projects to pursue and which to abandon, when to take on additional students and when to concentrate on current ones, and whether to spend a week in the lab or at a workshop. If they choose well, scientists increase their chances of living rewarding lives, producing useful results, and leaving a legacy of admiring students and colleagues ^ not to mention receiving tenure, grants, patents, and other tangible bene¢ts. When individual scientists choose poorly, they are the primary losers. Although funders might lament not having invested their resources more wisely, the opportunity costs of single projects are seldom that large. The stakes increase when investments are made in areas, rather than in individual projects. Sciences (and scientists) could do quite well for themselves, even as sponsors receive a poor return on substantial investments. This essay o¡ers a framework for managing the allocation of research resources, drawing upon the tools of decision theory (Clemen, 1991; Grayston, 1960; Green and Swets, 1966; Keeney, Hammond and Rai¡a, 1998; Morgan and Henrion, 1990; Rai¡a, 1968) and experiences with priority setting (National Research Council, 1995a,b; 1998a,b). 73 Policy Sciences 33: 73^87, 2000. ß 2000Kluwer Academic Publishers. Printed in the Netherlands.