Fox News is, by far, America’s dominant TV news channel; in the second quarter of 2017, Fox posted 2.35 million total viewers in primetime versus 1.64 million for MSNBC and 1.06 million for CNN. Given that Fox was founded by a longtime Republican Party operative and has almost exclusively hired conservative commentators, talk radio hosts, and the like to host its shows, it would stand to reason that its dominance on basic cable could influence how Americans vote, perhaps even tipping elections.
A new study in the American Economic Review (the discipline’s flagship journal), with an intriguing and persuasive methodology, finds exactly that. Emory University political scientist Gregory Martin and Stanford economist Ali Yurukoglu estimate that watching Fox News directly causes a substantial rightward shift in viewers’ attitudes, which translates into a significantly greater willingness to vote for Republican candidates.
They estimate that if Fox News hadn’t existed, the Republican presidential candidate’s share of the two-party vote would have been 3.59 points lower in 2004 and 6.34 points lower in 2008.
For context, that would’ve made John Kerry the 2004 popular vote winner, and turned Barack Obama’s 2008 victory into a landslide where he got 60 percent of the two-party vote.
“There is a non-trivial amount of uncertainty” about those estimates, Yurukoglu cautions. “I personally don’t think it’s totally implausible, but it is higher than I would have guessed prior to the research.” And even if the effect were half as large as estimated, that’d still mean that Fox News is having a very real, sizable effect on elections.
How Fox News transformed America
Martin and Yurukoglu integrated a vast array of data — on Fox’s channel position and viewership, individual/zip code/county level presidential voting behavior, and transcripts of cable news shows to showcase their ideology — into an extensive model that they can then use to estimate how effective Fox (and CNN and MSNBC) is at persuading viewers to vote its way.
Below, for instance, is how the estimated ideological stance of each channel changed over time; a lower score means more liberal, and a higher score means more conservative. You can see Fox News growing more conservative and MSNBC starting its move to the left with Keith Olbermann, Rachel Maddow, and others in the late 2000s:
Martin and Yurukoglu 2017
The effects of CNN and MSNBC on centrist voters are mostly negligible; MSNBC, in 2000 and 2004, modestly increased odds of voting Republican, before it turned left in time for 2008. But Fox News increases Republican voting odds for centrists, for Democratic viewers, and even, in 2004 and 2008, for Republicans already strongly inclined to vote that way. Watching three minutes more of Fox News per week in 2008 would have made the typical Democratic or centrist voter 1 percentage point likelier to vote Republican that year.
“Fox is substantially better at influencing Democrats than MSNBC is at influencing Republicans,” the authors find. While most Fox viewers are Republican, a sizable minority aren’t, and they’re particularly suggestible to the channel’s influence. In 2000, they estimate that 58 percent of Fox viewers who were initially Democrats changed to supporting the Republican candidate by the end of the election cycle; in 2004, the persuasion rate was 27 percent, and 28 percent in 2008. MSNBC, by contrast, only persuaded 8 percent of initial Republicans to vote Democratic in the 2008 cycle.
These are big effects, with major societal implications. The authors find that the Fox News effect translates into a 0.46 percentage point boost to the GOP vote share in the 2000 presidential race, a 3.59-point boost in 2004, and a 6.34-point boost in 2008; the boost increases as the channel’s viewership grew. This effect alone is large enough, they argue, to explain all the polarization in the US public’s political views from 2000 to 2008.
What’s more, they find that Fox isn’t setting its ideology where it ought to to maximize its viewership. It’s much more conservative than is optimal from that perspective. But it’s pretty close to the slant that would maximize its persuasive power: that would result in the largest rightward movement among viewers. CNN, by contrast, matched its political stances pretty closely to the viewer-maximizing point, showing less interest in operating as a political agent.
This is a “partial equilibrium” estimate: The estimates of Fox News’s effect are relative to a counterfactual where it disappears and only CNN and MSNBC remain. The authors are implicitly assuming another similarly conservative channel wouldn’t have emerged. “As a result, many (not most, but a substantial number) of former FNC viewers substitute into a relatively much less conservative option,” Martin explains. “In reality, you might imagine that a new conservative channel might have entered to replace FNC, or MSNBC or CNN might have moved right to capture the former FNC audience had FNC exited for some reason.”
But the result also jibes with existing research on the importance of Fox News. Studies looking at the initial rollout of Fox News in the 1990s found similar effects: There was increased support for GOP positions on controversial issues in places where Fox News was introduced, and increased GOP vote share too.
“Our estimates imply that Fox News convinced 3 to 28 percent of its viewers to vote Republican, depending on the audience measure,” economists Stefano DellaVigna (Berkeley) and Ethan Kaplan (Maryland) found in a seminal 2007 paper.
But, as political scientist Matt Grossmann (Michigan State) and David Hopkins (Boston College) have noted, these studies likely underestimated the effects of Fox, because they only looked at the 1996 to 2000 rollout of the channel, when it was much less watched. It would stand to reason that these effects would grow as the channel became more popular and more conservative.
That’s what makes the latest study so important. It builds on the prior research, confirms it, and shows how Fox’s increased popularity over the 2000s amplified its effects.
How the study works
Cable menus, saving us once again
The coolest thing about this research is the methodology. It’s really hard to estimate the effects of media outlets on individuals’ behavior, as media consumption is a two-way street. Yes, media can change peoples’ opinions and behavior, but people also choose to consume particular media because it aligns with their opinions and affirms stuff they’re doing already.
And prior economic research on media bias has found that media outlets’ political stances are demand-driven: that is, they take the positions they do because they want to gain readers/listeners/viewers. In this interpretation, Fox News might just be producing segments depicting food stamp recipients as lazy lobster-eating surfers because their audience already hates food stamps and welfare programs and wants something with which to agree. Confirmation bias is a hell of a drug.
So figuring out that a given media outlet is changing viewers’ minds, rather than merely reflecting their viewpoints back to them, is tricky. But Martin and Yurukoglu figured out an ingenious way around that problem: channel ordering.
It turns out that more people watch Fox News when it has a lower channel number. Fox News’s average channel number is around 38 to 41 (depending on which of Martin and Yurukoglu’s samples you’re looking at) and lowering the channel number to 19 to 23 or thereabouts causes viewers to watch 2.5 more minutes per week of Fox News, on average. In practice, that could translate into no effect on most people and a bigger effect (like, an hour more viewing per week) among a minority of cable subscribers — 2.5 minutes is just the overall figure.
What’s more, it doesn’t appear that cable or satellite TV providers make channel position decisions based on local politics; they don’t lower Fox News’ channel number in conservative towns or countries or raise it in liberal cities. So people in areas where Fox News has a low channel number watch more of the channel for reasons that are basically random, and unrelated to the viewer’s personal politics. That makes channel positioning a bit like a randomized experiment: Some people are randomly provoked to watch more Fox News than others, enabling researchers to see what effect watching Fox had on them.
Especially combined with the prior, also rigorous research looking at Fox’s initial rollout, Martin and Yurukoglu’s paper provides powerful evidence that Fox News is a critically important actor in American politics. It’s doing more than serving a market need; it’s actively reshaping American public opinion.
Vox · by Dylan Matthews