The present study reviews problems in the political learning literature, including ambiguous causality and a lack of specificity in linking communication content to learning outcomes. As a partial solution, our study of media and discussion influence incorporates both manipulated and observed aspects of mass and interpersonal communication. Results indicate that beyond (and often more important than) experimental manipulations, selection processes in news use and variations in the content of political discussions within exposure conditions matter for political knowledge. However, findings vary in predictable ways depending on the form of knowledge-overall factual knowledge, issue-specific knowledge, or knowledge structure density. These results suggest that the process of political learning via communication is more complex than it is often treated empirically.Keywords: Political Conversation, Deliberation, Political Knowledge, Structural Knowledge, News Use, Diversity, Talk. doi:10.1111/jcom.12138 Given that individuals' direct experience with major political events and policy debates are rather limited, researchers have assumed that news media use is a primary source of political information. Therefore, political communication scholars have devoted substantial research to confirm a positive relationship between news exposure and political knowledge. And very early in the history of research on political communication, studies demonstrated that personal interactions may also be relevant channels of political information. Reviews of the research suggest that news use and political discussion are important predictors of political knowledgeCorresponding author: William P. Eveland; e-mail: eveland.6@osu.edu (see Eveland & Garrett, 2014;Schmitt-Beck & Lup, 2013), and that they should be considered to work in tandem in many cases (Southwell & Yzer, 2007).Despite the valuable insights provided by prior scholarship on the effects of news use and political discussion on political knowledge, much of the research suffers from multiple limitations related to design and measurement (see Slater, 2004;Southwell & Yzer, 2007). We begin by reviewing the evidence on communication's relationship with political knowledge, paying particular attention to the methodological limitations of typical political knowledge studies. We then introduce this study, whose design addresses some of the problems with prior research by employing experimental manipulations of news exposure and public affairs discussion in small group settings in which some members have, and some have not, had exposure to relevant news prior to discussion. Just as importantly, the study goes beyond standard experimental and survey procedures by incorporating detailed observational data-content analysis of face-to-face conversation transcripts and computer tracking of online news use-to provide a more precise understanding of the specific aspects of news use and discussion that produce both content and structural aspects of public affairs knowledge. Thus, as advoca...