From anonymous online commenters up to the highest levels of American leadership, public discourse is awash today in accusations that the media are politically slanted. But to what extent is that true and how does it affect politics?

To explore those questions, Knowable Magazine turned to political economist David Strömberg of Stockholm University, who researches media influence on politics and policy. Strömberg examined media bias around the globe and through history in a 2015 article in the Annual Review of Economics. His findings are counterintuitive, revealing that ideological bias may have less impact than we expect. This interview has been edited for length and clarity.

How long have researchers studied media bias?

Researchers started thinking about these things because of the rise of fascism in Europe in the 1930s, which was also around the time radio was introduced. People saw that the Nazis and fascists in Europe were using radio a lot for propaganda. This raised fears from onlooking nations that their own citizens could be brainwashed with propaganda. And so the first studies set out to test that.

Some early research interviewed U.S. voters before a presidential election, looked at the news material they consumed, and asked whether their opinions shifted based on whatever news material they had been exposed to. They found that people’s ideological or political standpoints rarely changed. The reason for that was that Republicans were reading Republican newspapers and Democrats were reading Democratic papers, so people were basically just becoming more firm in whatever belief they had before.

How do the media function when everything is working?

The ideal role of the media is to provide information. It’s almost impossible for people to inform themselves about what politicians are doing directly, and it’s just boring to many people. Media is a very effective way of having a few big actors with lots of resources finding stories of interest and presenting them to voters. This makes politicians much more accountable.

And what does damaging bias look like?

The way that economists have defined damaging bias, it typically means that you’re misrepresenting the facts, suppressing facts or just lying. This would be different than, let’s say, a conservative newspaper endorsing a Republican candidate — this kind of ideological bias is not necessarily suppressing any information or lying.

Today we’re hearing a lot of concern that many media outlets in the U.S. are ideologically biased. What does research tell us — are these concerns well-founded?

Researchers look at things like what candidates and ballot propositions newspapers endorse and what kind of think tanks they cite, and then compare newspapers to members of Congress and Supreme Court judges. There has been an increase in the polarization of U.S. media, to the left and right. But if you look at empirical studies of how biased the U.S. media are, on average it’s been pretty close to centrist.

These studies look at a large set of U.S. newspapers. There are, of course, extreme cases where outlets are ideologically biased. But if you’re thinking about the average influence of the media, you should consider the average position of the media.

What about more implicit measures, like a 2014 survey that found that, while most American journalists identified as independent in 2013, four times as many identified as liberal than identified as conservative?

There’s been a long discussion of slants of different important actors in the newsmaking process. Typically, journalists are left of center, whereas advertisers and owners may be more right of center. Of course, the end output that we care about is what is in the newspapers. Typically, studies find that nowadays in the U.S., newspapers are slightly to the left of center, but still moderate compared to other actors like politicians and Supreme Court judges.

A newspaper's front page shows a drawing of the U.S.S. Maine exploding.

On February 17, 1898, Joseph Pulitzer’s paper, The New York World, featured the unverified claim that the explosion of the U.S.S. Maine was the result of enemy fire. William Randolph Hearst ran similarly dubious claims of “Spanish treachery” in his competing paper, The New York Journal. The heated rivalry between the two papers fueled the production of sensational, and sometimes disreputable, news — and led to the term “Yellow Journalism.” Strӧmberg characterizes such misrepresentation of facts as a damaging form of media bias.


Is this the first time the U.S. has been dealing with media bias?

No. If you go back to the late 1800s, most newspapers had a party affiliation and their content was much more slanted. Around the 1890s to 1920s, the share of newspapers who had strong party affiliations dropped drastically, though even then there were newspapers with some Republican or Democratic leaning.

What caused the shift away from explicit party affiliations?

People have speculated. There is some evidence that advertising revenue and competition between papers may be what was driving the change. As it became more profitable to write stories that interested people, biases fell. One study found that in areas where ad revenues were higher, the slant was less.

And since then, it’s been standard for media outlets to strive for balance in their coverage?

Actually, that’s really an American norm. In Europe, that shift didn’t happen — most newspapers have a clear political affiliation. But that doesn’t mean that they were incredibly slanted — certainly not biased in the sense of producing fake news. It just means that they have a standpoint that’s transparent.

When a news outlet has an ideological bias, how much power does it have to carry that over to its audience?

It’s always been the dream of researchers to find big effects of media on how people are voting. I mean these are the hypotheses the studies in the ’40s started with. But one of the consistent findings all the way back to the first big studies is that it’s very difficult to change people’s voting intentions — for example, what political party they prefer. If you are a right-wing person and you get left-wing media, or vice versa — first of all, you just don’t read it. And even if you were exposed to some news that is not aligned with your ideological positions, you just wouldn’t take it in.

But that doesn’t necessarily mean it doesn’t affect election results, because it could be that the media increase voter turnout by energizing voters.

You argue in your review article that this tendency of voters to seek out media that are aligned with their own biases might, counterintuitively, be good for voters. How?

Political accountability works best if voters don’t make mistakes. By mistake, I mean voting for a candidate that is not the better candidate for you. If you have a media source that has the same exact preference as you do, and you just follow their endorsement, for example, you would never make a mistake.

So it can benefit voters to seek out news sources that share their views. But are there drawbacks to that as well?

One drawback is that ideological media outlets polarize the electorate by reaffirming voters’ pre-held positions. This means that the media are likely to make the Democratic voters support Democrats more strongly and the Republican voters support the Republicans more strongly.

And a very polarized electorate is not good for political accountability, because strongly partisan voters will not be very responsive to important nonpartisan factors. A politician might be slightly corrupt or not be very efficient in policymaking, but very few voters care — because they only care about his strong ideological positions on, let’s say, abortion.

What happens when slant crosses into damaging bias — not just ideological, but misrepresenting facts, suppressing facts or lying? Can the media harm society?

Some evidence exists that the media could provide worse outcomes for society, but this comes mainly from totalitarian states where there’s little media competition — there’s just one media outlet and the rulers use it to implement terrible actions. One example was how radio was used by the Rwandan government in the 1990s to, basically, tell people to commit genocide. One study found that where there was better radio reception, there were more civilian deaths. Another study from 2013 shows that the Nazi control over radio broadcasts increased support for anti-Semitic policies in places where there was a prior history of rioting and attacks targeting Jewish communities. But these are extreme examples that don’t tend to happen in democracies with a free press.

What do negative media effects look like in democracies with a free press?

There are instances of people being misguided by the media. For example, in the run-up to the Iraq War, much of the mainstream U.S. media coverage uncritically supported the government line that Iraq was developing weapons of mass destruction. Outlets like the New York Times later apologized for mistakes in their reporting.

“Because we are emotionally outraged by media bias, we tend to think about it a lot and forget about newspaper deaths and journalist layoffs.”

David Strömberg

But the negative effects of media in democracies usually have more to do with influencing politicians to focus on the “wrong” issues. Obviously, some issues are intrinsically more newsworthy, like plane crashes or volcanic eruptions as opposed to events like traffic accidents, famines or endemic hunger, which are just constant problems — not “news.” News coverage puts a spotlight on certain issues and incentivizes politicians to work on them.

In democracies with a free press, what should people do if they’re worried their news is biased?

It’s like with any other business: Consumers apply pressure by what they consume. So you have to switch to a different media source if you don’t like what another source is doing.

How does media bias compare to other issues facing journalism today?

Because we are emotionally outraged by media bias, we tend to think about it a lot and forget about newspaper deaths and journalist layoffs. But there is much less evidence that media bias matters significantly. There is quite a bit more evidence that the volume of news matters more for making democracy work.

The media market in the U.S. is very mature, and there are many newspapers. And even consumers of the most biased media news sites often consume other news. So in a media market like the U.S., if you add one additional news source with a Republican or Democratic slant to the many others, it makes little difference.

And even newspapers with strong Republican or Democratic news slants are typically still covering their in-party corruption scandals. So it’s much more important that you have a local newspaper than if it has slant one way or the other.

Of course, if you go someplace like Russia where they don’t have many media outlets, then media slant matters much more. If you go from zero opposition television stations to one, it makes a huge difference.

And a higher volume of financially viable presses means the press is harder to silence.

How so?

A theoretical argument is that you have to silence each and every outlet in order for information not to get out. You can shut down outlets and throw journalists in jail — like we’ve seen happening in Turkey — or bribe the press, or hire editors more politically aligned with leadership. But if even one news outlet covers a piece of information, then the news will be out.

A clear example is a paper that looked at bribery by leadership in Peru. In the 1990s, Peru’s secret police chief, Vladimiro Montesinos, paid out many bribes and he recorded it all — all the bribes paid to different branches of government and the press to keep them silent about corruption. One important finding in that paper is that it was much more costly to silence the media than to silence judges and legislators. Typical bribes to TV channels were about 100 times more than bribes to politicians.

Strömberg says that maintaining political accountability is an important role of the news media. Yet some survey respondents said they felt that criticism in the news can keep politicians from doing their jobs. This sentiment tends to flip-flop along party lines, depending on who is in power.


You wrote your review in 2015. Have there been important changes since then?

There are two big ones — falling advertisement revenues and social media. I think social media is changing the media landscape both in countries like the U.S., but even more so in countries like China that don’t have a mature and independent press and media market. It’s much more difficult to censor millions of users than it is to censor a few media outlets.

But leaders in dictatorial regimes can also use social media for surveillance, paying firms to look for posts that predict protests and other similar events before they happen. This is a big industry in China, where calls for protests have been picked up by politicians, and then they take measures like making work days on the weekend, and forcing students to be in school, to prevent a protest.

And what about the spread of false information on social media?

There’s two obvious reasons why you see this kind of fake news on social media. One is that those who post fake news want to influence an election, or something else. The other is that it’s profitable. We have these fake news-producing sites in Macedonia who know nothing about U.S. politics but they know how to write a news article or post that will get many clicks. The thing that makes social media conducive to fake news is there’s much less of a reputational concern for users on Facebook or Twitter. They can write something crazy and then just open up another account, and write something crazy. A media outlet in the U.S., like a local newspaper, cannot do that. Reputation is what makes people pay for their articles.

And what is the impact of falling ad revenues?

Falling advertisement revenues is a major issue now in the media industry, leading to a less financially viable press. So we have to think about if we need to start having government subsidize the media. There’s an economic argument for this, based in economic theories about behaviors that affect other people. If you are better informed, then the politicians will become less corrupt — this is good for you and good for everyone else. Just as you might put a tax on something that is bad, like polluting, you want to have a subsidy on something that is good, like getting informed.

But the effects of government subsidies on bias are not clear. It’s a topic for future research.