The alleged misuse of scientific information to support public policy decisions has become one of the most prominent and controversial topics in American politics. Perceived government misuse of scientific data in highly controversial areas such as global warming, environmental protection, stem cell research, and contraception threatens to lead us not only toward policy positions that are inconsistent with scientific reality, but perhaps more importantly into a political environment where science-based policy decisions are no longer viewed as legitimate within our constitutional democracy. Due in part to these concerns about scientific integrity in administrative decisions, there is also a significant amount of attention being paid to scientific peer review in the administrative process ("administrative peer review"), a movement highlighted by a 2005 OMB bulletin mandating that administrative agencies obtain peer review of all important scientific information that they disseminate to the public.

This article addresses the cross-section of these two issues through a normative analysis of peer review's impact on administrative legitimacy. To date, commentary on peer review in the administrative context has been limited to treating peer review either descriptively or as a largely monolithic enterprise. This article departs from that approach by asking the next logical questions - what model(s) of peer review are available to administrative agencies, and which of these is best suited to promote legitimacy in administrative decisions - and by creating a framework for answering them, first through the development of four distinct models of administrative peer review, and then by establishing a series of normative approaches, including a cost-benefit analysis, through which to evaluate each of the models. In addition to providing a new theoretical context in which to consider peer review, the results of the analyses are interesting in that they overwhelmingly support the multifaceted treatment of peer review developed here and raise serious questions about the most commonly used model of administrative peer review.