New media has changed people’s experience with news. News readers nowadays encounter both selective opinions from elite sources and comments from anonymous strangers. The question is: how do people simultaneously process these two types of information? This dissertation selects a health policy, namely the cigarette graphic warning label (GWL) policy, locates online news reports on the major developments of the GWL policy, examines the content and dynamics of the public deliberation on the comment boards for these news articles, and explores the social consequences of such deliberation on news readers. A computerized content analysis was first conducted on user-generated comments following GWL news articles and results showed the majority of the comments were relevant to the issue under debate and argumentative and thus qualified as public deliberation. Comments were predominantly against GWL, and the most prevalent argument was the danger of government infringing on personal life. Three thematic frames emerged from the coding of arguments in comments: the legitimacy of the policy, the effectiveness of the GWL, and the presentational features of the labels. An experiment was then conducted to test the effect of news and comments on readers’ attitude and behavior. Readers of oppositional comments showed significantly lower level of policy support than those who read no comment or supportive comments. News story elicited the highest level of policy support when only the basic facts of the policy but none of the argumentative themes was covered. Comments outperformed news in shaping readers’ thought diversity such that comments could stimulate people to think more when news is narrow, and limit people to think less when news is thorough. Political ideology interacted with comment valence to influence participation such that conservatives tend to post comments if the opinion climate is overly positive, but liberals did not show interest in posting when the opinion climate is overly negative. Comments are a distorted reflection of public opinion. Content analysis found only 10% of the comments expressed any form of support for the GWL policy while 61% of the experiment participants indicated they were in favor of the policy.