Partisan polarization has been at the center of scholarly attention. Many studies on this topic have underscored elite polarization as a primary cause of mass polarization in American politics. Relatively little attention has been paid to the role of mass media in promoting polarization. This is partly because of the notion that news media largely reflect political reality. However, when it comes to news making, the role of journalists should not be underestimated: Journalists choose what to report among thousands events happening all over the world and how to report the event. This active role of journalists in the news reporting process is conceptualized as framing. Conflict is known as the most prevalent way American journalists frame their stories (de Vreese, 2012; Neuman, Just, & Crigler, 1992). In this present dissertation, I argue that conflict framing—especially news media emphasis on partisan conflict in their coverage of political events—contributes to partisan polarization. Drawing on self-categorization theory of group polarization (Turner et al., 1987), I hypothesize that as a vicarious experience of partisan conflict, conflict-framed news will lead people to respond to the disputed issue as members of the Democratic or Republican Party instead of as unique individuals; this increased partisan self-categorization is a mechanism behind the news effect. This group polarization hypothesis is tested with three experimental studies. Specifically, with a use of diverse samples of students and non-students, Studies 1 and 2 find supporting evidence that Democrats and Republicans exposed to partisan conflict-framed news adopt more polarized positions along party lines when news exposure elicits their partisan identity. To rule out a potential confounding effect of the so-called party cue effects (Goren, Federico, & Kittilson, 2009), Study 3 retests the group polarization hypothesis in the context of gender polarization. The findings of Study 3 are robust. Exposure to gender conflict-framed news promotes attitude polarization between women and men; increased gender self-categorization accounts for the news effect. The findings inform us that media reflection of elite polarization (Cohen, 2003; Druckman, Peterson, & Slothuus, 2013) and the growing popularity of partisan cable news channels (Bennett & Iyengar, 2008; Stroud, 2010) fail to paint a whole picture when it comes to the link between news media and polarization in American mass public. Many scholars have criticized the news media for transforming politics into a “strategic game” (Patterson, 1993) or “spectator sport” (Price, 2009) and are concerned about the potentially-negative influence of this style of coverage on deliberative democracy. By examining the polarizing effect of news coverage that oversells the competitive aspect of party politics, the present dissertation study provides some of the only evidence that speaks to these claims. The second part of this dissertation examines whether self-categorization theory sheds new light on polarization research beyond and above current literature. To that end, I explore three possible mechanisms through which conflict framing may elicit group polarization of attitudes: (1) by increasing group self-categorization, (2) by increasing motivated reasoning, and (3) by increasing intergroup animus. Data from Studies 1, 2, and 3 provide promising but conflicting results: For partisan polarization, motivated reasoning is the most powerful explanatory mechanism behind the polarizing effect of the news. Yet, for gender polarization, self-categorization theory was the most powerful mechanism. Understanding of how conflict framing promotes partisan polarization is critical because it paves the ways to combat the negative effects of the news. Implications and directions for future studies are discussed.