将受众公民的兴趣考虑在内。因此新闻议程设置也是关于熔合公众兴趣的理论。
要理解人们如何组成他们自己的社区议程，两个简单而重要的概念是垂直（vertical）和水平（horizontal）媒体。垂直媒体指报纸、广播和电视这些传统媒体，它们自上而下地接触受众，就像是从高山上发出呐喊。而水平媒体则是满足个体兴趣，就像杂志和社交网络那样的媒体。它们有着各自独特的受众，将世界上一个个单独的个体链接在一起。当今的世界正在目睹水平媒体设置议程力量的增强，社交网络平台将传统的垂直媒体转化为满足个人所需的水平传媒。与此同时，社区议程则被逐步转化为个体议程。
例如，一个2008年对1968年教堂山研究的复制研究发现选民熔合了垂直和水平议程，他们从垂直媒体获取一些信息，从水平媒体中获取另外一些，然后将两部分熔合成为自己关心的议程。在这个过程中，对垂直和水平媒体的选择取决于个体对选举话题的兴趣。如果我们假设一个封闭的、由直接经验和间接来源组成的信息系统，个体的议程可以由下面的公式来预知：
假设垂直媒体议程和选民个体关心的议程之间的关联度是0.8，那么未解释的部分，即来自水平媒体的影响是0.2.将这些数值平方再从整体1之间减去，得到的是选民自己控制的部分（0.32），这部分就是垂直和水平媒体都未能解释的部分。这个公式可以用来通过调查媒体议程，来预测选民关心的话题议程。
基于这个公式，肖用一张线性图描绘了议程熔合的过程。这张图的左侧描绘了一个相对稳定的传统社会模式，垂直媒体和公共议程之间的关联很高。中间部分是传统的整体的社区向分散的社区转化的过程，而当垂直媒体议程和个体议程关联度降到0.5以下，则是此图右侧所描绘的分散的社区替代传统社区的情形。肖说，目前这个假设还有待在世界范围内收集现实数据加以验证。
最后，肖谈到了当代媒体景观和大众媒体受众的演变。随着传统的垂直媒体受众规模逐渐缩小，垂直媒体的议题设置效果可能也在变小。与此同时，个体在选择使用和熔合媒体议程的方式上会呈现代际的差别。这些新特点意味着研究者需要付出更多的努力来理解当今的受众个体独特的议程混合方式，这些努力也将帮助政府和公民来理解一个迎合个体兴趣的水平媒体占支配地位的社会。
三位理论创始人总结了议程设置理论三个重要和特别活跃的研究领域：网络议程设置，导向需求和议程熔合。议程设置理论即将走过50年的历程，如今已经成为一个相当成熟的理论，其每一个分支都在引领着现在和将来的实证研究。麦库姆斯、韦弗和肖期待未来的议程设置研究可以突破以往对公共事务的关注，在延展到其他话题领域的同时加深人们对议程设置理论核心理念的理解。
三位学者对年轻学者的忠告是尽早建立自己的研究方向。他们三人的成就则是活生生的成功典型。他们整个职业生涯持续地为议程设置理论奉献精力，如今退休后仍在努力开创新的研究领域，他们的研究已经成为全世界议程设置理论和实证研究不可或缺的参考文献。
“Three Big Heads” on the future of agenda setting
The three founding fathers of Agenda-Setting theory, Drs. Donald Shaw, Maxwell McCombs and David Weaver, presented a colloquium at Reese Phifer Hall at the University of Alabama on November 14, 2014. Students and faculty of the college of Communication and Information Sciences heard first hand from the ―three big heads‖ about the theoretical implications and the latest applications of agenda-setting theory in today’s digital and fragmented media.
McCombs and Shaw are known to write the first agenda-setting paper. They proposed the theory in their study of the 1968 presidential election in Chapel Hill, North Carolina. Since its publication, the 1972 article ―The Agenda-Setting Function of Mass Media‖ has been one of the top 10 most- cited articles in mass communication research every year. In 2011, the World Association for Public Opinion Research recognized McCombs and Shaw with the Helen Dinerman Award and commended them for influencing ―a paradigm shift in how we think about the media and public opinion.‖
Weaver is well known for adapting and examining the concept of indi- vidual’s need for orientation (NFO). His work created a second theoretical area
for agenda setting and provided a psychological ground for explaining the agenda-setting process. Weaver’s recent endeavor is tracing the origins of media agenda, which links agenda-setting research to another area of mass communication, the sociology of news.
McCombs retired from the faculty of the School of Journalism at the University of Texas at Austin in 2011 as the Jesse H. Jones Centennial Chair in Communication. Shaw, also a professor emeritus, still holds the Kenan Professor in the School of Journalism and Mass Communication at UNC-CH. Weaver is Roy Howard Professor Emeritus after more than 37 years of teaching and research at Indiana University.
The ―three big heads‖ talked about different avenues and domains of agenda-setting research. However, each of their presentations touched the frontiers of the agenda-setting research, respectively, on the third-level agenda setting (McCombs), need for orientation (Weaver) and agenda melding (Shaw).
McCombs: Power of information networks
McCombs has focused specifically on the ―third level‖ of agenda set- ting, or network agenda setting, a new area of agenda-setting research based on decades of studies of first- and second-level agenda-setting effects.
The first-level of agenda setting traces back to McCombs and Shaw’s (1972) original hypothesis that the mass media’s agenda for political campaign would influence the salience of the political issues in the mind of undecided voters. It examines the correlation between the salience of news subjects and perceived key issues among the voters. The second-level agenda setting was proposed soon after in Shaw and McCombs’s (1977) follow-up study of their first agenda-setting study, theorizing that not only the salience of news subjects would transfer to the public’s perception, but also the attributes’ salience of news. That is, the characteristics of the subject as well as the affective components associated with these subjects, such as tones (positive, negative or neutral) of coverage, would affect how the public perceive the news subject. The third-level agenda setting, according to McCombs, has to do with ―network agenda setting.‖
Compared with the first- and second-level agenda-setting research that assume a discrete transfer of the salience of issues or attributes from the news media to the public, the third-level agenda setting theorizes that the subjects and attributes’ salience transfer in bundles between agendas. In other words, the news media not only tell us ―what to think about‖ (first-level agenda-setting effect) and ―how to think about it‖ (second-level agenda-setting effect), but also tell us ―what and how to associate.‖
McCombs attributed the theoretical ground of third-level agenda setting to Kaplan’s (1973) and Anderson’s (1983) associative network model of memory. Call it the ―cognitive maps‖ (Kaplan, 1973) or ―architecture of cognition‖ (Anderson, 1983), the network model of memory asserts that there is a general tendency for individuals to associate different elements in their minds as an effort to make sense of social realities. To form what Walter Lippmann called ―picture in our heads,‖ individuals turn to media to experience public events and absorb what they are exposed to. At the end of the day, the ―pictures in our heads‖ is a gestalt of elements linked in a variety of patterns. The third-level agenda setting, at its core, is an exten- sion of this gestalt perspective of agenda-setting theory.
There is empirical evidence for third-level agenda-setting effects. One initial effort was the reanalysis of a second-level study of Kim & Mc- Combs’s (2007) examining candidate images among voters and in their local newspaper. A network analysis revealed a correlation of .67, which is very close to the original attribute agenda-setting correlation at .65. Some new