Scientific information, used by the U.S. federal government to formulate public policy in many arenas, is frequently contested and sometimes altered, blocked from publication, deleted from reports, or restricted in some way. This dissertation examines how and why restricted access to science policy (RASP) occurs through a comparative case study. Two frameworks are typically used to analyze restrictions of science policy. Principal Agent Theory (PAT) examines circumstances in which principals employ agents. Democratic accountability explains the value of circulating information among bureaucrats, scientists, and the general public and the dangers of restricting access.

Four cases, based on semi-structured interviews with scientists, bureaucrats, and other key actors, were analyzed. The interviews were recorded, transcribed, and analyzed with QSR NVIVO-8. Comparative analysis yielded several themes: the nature of the information in question, the role of that information, and the intended audiences. Regarding the nature of the information, multiple respondents emphasized the rigor and objectivity of science, yet noted the uncertainty inherent in research, particularly complex topics such as climate change. Respondents suggested multiple roles for science policy: pure information, a basis for policy decisions, and one of several policy inputs. Though they did not use the language of PAT, policymakers indicated they were accountable to multiple principals. Finally, many respondents viewed the general public as an important audience for science policy, particularly because democratic power is invested in citizens; at the same time, the public is often uneducated or uninterested in the intricacies of science policy, potentially negating the theory of democratic accountability. The mechanics of RASP vary; while individual actions can be identified, RASP does not consist solely of these discrete actions, but of the broader context in which they take place. Within that broader context, respondents explained that RASP occurred for ideological or economic reasons. Additional research is recommended to determine the applicability of these conclusions to other instances of information restriction.