The conventional wisdom about mass-media effects has long been that media might set the agenda or frame issues in a particular way, but that they typically have little direct persuasive effect on attitudes, tending primarily to reinforce existing beliefs. But the return of partisan media outlets, particularly on cable television and the Internet, has led scholars to question whether that conventional wisdom needs to be updated. In particular, some have argued that partisan outlets contribute to political polarization in the mass public. Kevin Arceneaux and Martin Johnson, however, argue that such fears are largely unfounded. Looking primarily at cable news (Fox News and MSNBC), they show that partisan media viewers come to these shows already polarized, and the programs do little to drive them farther to the extremes. Indeed, looking across a variety of different outcome variables, they argue that partisan media outlets have only a very limited effect on their audience. The larger and potentially more pernicious effect, they claim, is that viewers in the contemporary high-choice media environment can simply choose to opt out of news altogether, echoing Markus Prior’s (2007) claim in Post-Broadcast Democracy.

Arceneaux and Johnson develop a novel theory to identify the individuals who actually tune in to partisan media programs, which they label the active audience theory. They argue that there are two relevant motivational factors that influence the decision to watch partisan media programs: a partisan motivation and a news (vs. entertainment) motivation (see especially Figure 3.2 on page 52). Only those who prefer news and have strong partisan motivations seek out partisan news content. Because these individuals are already more partisan and ideologically extreme before tuning into these programs (indeed, their partisan and ideological proclivities are what drives them to tune in in the first place), there is a limit to the influence that partisan news outlets can have. While the authors do not put it in these terms, their argument parallels the “pretreatment” logic spelled out by James Druckman and Thomas Leeper (“Learning More from Political Communication Experiments: Pretreatment and Its Effects,” 2012): These individuals have already encountered the extreme rhetoric on these programs in other places, and so the shows themselves have a limited effect.

The authors use an extremely novel set of experiments to test this general theoretical claim. In their design, only some subjects are forced to watch partisan media programs (as in a standard design), while others are allowed to choose their preferred program by using a TV remote to choose among partisan news and several entertainment options. This choice condition, then, allows them to estimate the effects of motivation: What happens when some subjects can opt out of partisan media programs, as they can in the real world? They find that in this choice condition, many of the previously found effects of partisan media disappear. They further show that what effects do persist are primarily concentrated among those who prefer entertainment to news, precisely those least likely to tune into these programs. This suggests a rather limited political effect of partisan outlets.

Such carefully constructed and innovative experiments make the book required reading for anyone working in political communication, and it will undoubtedly receive a great deal of well-deserved praise. That said, however, I want to raise two related issues where I am less persuaded by their results. I raise these issues not to be cantankerous but, rather, to suggest areas for future scholarship.

First, there is the question of what these experiments actually estimate. With a broad brush, their effects capture something akin to the population average treatment effect (What is the overall effect of partisan media exposure throughout the population?). By allowing for self-selection, they correctly illustrate that across the population as a whole, the effects of partisan media are quite modest, precisely because so few viewers choose to tune in to these shows. But it is unclear that this is the key quantity of interest. While the effect throughout the population is small, a small but active audience can have large political effects. As their own results show, these people are more partisan, ideological, and politically involved. Such individuals are more likely to make their voices heard, and they would also be more likely to be the opinion leaders who influence others through two-step communication flows (going back to the finding in Bernard Berelson, Paul Lazarsfeld, and William McPhee’s 1954 classic, Voting). In particular, this sets up the possibility of the indirect influence of these shows (to be fair, the authors acknowledge this latter point about indirect influence later in the conclusion). Further, such programs may well have important effects on elites or the broader political agenda. In sum, a small audience need not imply small effects. To me, this suggests that a conclusion of minimal effects is perhaps a bit premature.

Second, and more importantly, I worry that the issues used as experimental stimuli partially drive these null effects. Many of the experiments in the book use highly salient and well-known issues in American politics. In particular, in Chapters 4 and 5, the studies focus primarily on Barack Obama’s handling of the economy, the fate of health-care reform, and debates over tax increases, especially on the rich. These are arguably three of the major debates not only from Obama’s first term but from the past 50 years as well; subjects no doubt enter with strong priors on these topics, with little room for the segments to have an effect except for those with the weakest opinions. But what would happen for less salient issues? As Larry Bartels pointed out 20 years ago (“Messages Received: The Political Impact of Media Exposure,” 1993), media effects are likely concentrated on lower-salience and novel issues. On more salient issues, like the president’s handling of the economy, health-care reform, and taxes, their audience absorbed the effects of partisan outlets long ago, and an additional viewing will likely have no discernable effects (this the “pretreatment” point). On smaller, more emerging issues, however, partisan outlets might have more significant effects. This might help to explain why Arceneaux and Johnson find no effects, when a long line of studies in the motivated reasoning tradition would predict stronger effects. That said, however, in fairness to these authors, these programs do focus in large part (though certainly not exclusively) on these salient issues, and so the effects on these issues are important. Future work will be needed to explain how issue salience conditions the effects of these shows.

Such criticisms aside, Arceneaux and Johnson have written an impressive book that will draw the attention of a wide range of scholars in political communication, political psychology, and American politics more broadly. Changing Minds or Changing Channels? discusses a wide variety of phenomena, too many to adequately review in such a short space. In particular, it is a large step forward in the literature on self-selection into political media, and sets a new bar for work in this area. Indeed, I expect that the long-term effect will be to spur new work on the political consequences of biased media, and will be seen as a foundational work in this field. This book will generate a great deal of attention and debate, and will garner praise from a large number of scholars and readers alike.