Hillary Clinton speaks to Pennsylvania voters at Temple University in Philadelphia on Sept. 19. (Melina Mara/The Washington Post)

Do Americans have an insatiable appetite for scandal? It may seem so, given the amount of media coverage dedicated to controversies about Hillary Clinton’s email practices and Donald Trump’s comments about President Obama’s place of birth. But new evidence suggests otherwise: Our attention to scandal is extremely fleeting – and filtered by partisanship.

Here’s how I researched this

To research this, I turned to an online panel of Americans, using special software that runs in the background of their computers and records the websites these respondents visit. The members of this panel had previously agreed to install this software for research purposes. This offers a rare peek into how real people’s browsing habits change in response to controversies. (A note on the respondents: They are members of an online panel maintained by the polling firm YouGov. The ones in my data had already opted in to allow anonymous tracking of their web visits using software that can be uninstalled at any time.)

As it happens, data collection took place between Feb. 27 and March 19, 2015. The New York Times broke the email server story on the fourth day of data collection, March 2, offering a unique opportunity to test some theories about the causes and consequences of scandal coverage.

One theory is that the news media cover scandals because that’s what their audiences want. A countering theory suggests that what gets framed as a “scandal” is decided, at least in part, by the news media, which are paying attention to such things as which party is in power and what other stories are on the news agenda. Because most people aren’t paying close attention to politics, the news media decide what gets played as an important story and what gets ignored. Think of these as the “bottom-up” and “top-down” perspectives. A third theory suggests that partisan readers and viewers are hungry for scandals about the other side: Democrats will be more likely to pay attention to scandals involving Republicans, and vice versa.

To see how their media choices changed as a result of the story, I assessed the web visit histories with a measure of ideological slant. I used measures estimated by researchers at Facebook of the average ideological self-placement of users who share content from a particular domain. For example, most people probably would agree that Foxnews.com, the website for Fox News Channel, is more conservative than Huffingtonpost.com. But by how much?

According to Facebook users’ self-assessments, sharing articles from Foxnews.com corresponds to a self-assessment of about 0.78 on a scale of -1, for extremely liberal, to 1, for extremely conservative. Sharing articles from Huffingtonpost.com is almost but not quite as far from the center in the other direction, at -0.62. Very close to the ideological center are large news portals like msn.com (-0.08).

Here’s what I found in the data.

Respondents to the survey had already told me whether they self-identified as Democrats, Republicans or independents. Below, you can see the average ideological slant of all news-related media consumption for those identified along each of those lines, plotted by day:

Average ideological slant of the media consumption recorded each day by self-described Democrats, Republicans and independents (higher is more conservative). The Clinton email story was published online on March 2, 2015.
Data: YouGov; Figure: Andy Guess


Only Republicans changed their web habits when the private email server was revealed

As you can see immediately, after the email server revelation, Republicans – and only Republicans – immediately spent far more time on conservative news sites. (Because the respondents in my data were not recruited using probability sampling, I’ve weighted these results to be representative of national demographic characteristics.) That wasn’t true for Democrats or independents, whose media habits stayed remarkably stable for the entire three weeks.

Overall, Republicans in the sample visit somewhat more conservative websites on average (about 0.1 on a scale from -1 to 1), and Democrats visit somewhat more liberal websites (about -0.1). Independents’ visits are right in the middle. And of course, they’re dwarfed when there’s a scandal affecting the presumed Democratic presidential candidate, and Republicans’ media consumption suddenly surges to 0.6 on the conservative scale.

This pattern is consistent with the theory that people’s partisan identities tell them which scandals matter. In this case, part of the mainstream media, he New York Times, broke a story that was then covered thoroughly in conservative outlets, both online and off. These outlets broadcast a loud and consistent message that the email revelations were damning evidence about Clinton’s lack of fitness for office, thus signaling to readers and viewers that the scandal was important. But the signals were consumed only by those already receptive to that point of view.

Even among Republicans, only the most conservative were interested in the email server – and only briefly

Below, you can see that not every Republican was suddenly consumed by interest in Clinton’s email server. Only those who describe themselves in the survey as “very conservative” flocked to conservative news sites. Even most Republicans did not change their media consumption as a result of the scandal, just the most ideological.

Average ideological slant of the media consumption recorded each day by self-described ideological groupings (higher is more conservative). The Clinton email story was published online on March 2, 2015.
Data: YouGov; Figure: Andy Guess

This overall pattern is hard to square with the theory that readers are always eager for scandal coverage. Apparently the most partisan are happy to have scandal coverage that reinforces their existing beliefs. But even that subgroup quickly loses interest; strongly conservative Republicans returned rapidly to their previous pattern of media consumption.

Why didn’t Clinton’s emails hold public attention for long?

Here’s why: Most people pay relatively little attention to politics. It’s difficult for political events to break through into those limited windows. The ones that do tend to be major periodic occurrences (such as Election Day) or unexpected developments. When we respond, we seek out information that helps us assimilate these new developments into our preexisting worldviews or narratives. Then we move on with our busy lives.

To be clear, I’m not claiming anything particular about the way that Republicans or conservatives respond to political scandals. It just so happened that the scandal that broke in the midst of data collection was about a Democrat. I would expect the reverse pattern if the controversy were about a Republican.

But the Clinton email story tells us most people aren’t constantly demanding scandalous coverage of the people they consider to be political opponents. Instead, media coverage – including partisan media coverage – tells us when it’s time to pay attention.

Andy Guess is a postdoctoral researcher at the Social Media and Political Participation Lab at New York University.