This dissertation examines how information is translated into issue

definitions. Issue definitions---the way that policy issues are

understood---have long been noted to be important for policy

choices. In this project, I develop a model of issue definitions where

issues are understood as a function of the various dimensions of the

issue weighted by the importance of each dimension. I then incorporate

this model into the theory of information processing developed by

Jones and Baumgartner (2005). The theory of information processing

posits that information can be understood as signals in the

policymaking environment, and information processing is the collection

and prioritizing of those signals. In this dissertation, I model

these information signals as the salience of each dimension of an

issue.

Using the case of used nuclear fuel (UNF) management, this dissertation test

hypotheses about the nature of issue definitions and policy change,

institutions, and policy actors. Specifically, I estimate the

dimensions of the UNF issue using latent Dirichlet allocation, a type

of quantitative text analysis. Following the development of the UNF

dimensions, I test hypotheses about how the salience of these dimensions are related

to policy change, how institutional structures influence dimension

salience, and how policy actors systematically highlight some

dimensions over others.