Media Bias

The comments in response to my recent post on Barack Obama’s book led, predictably, to the topic of media bias.

Media bias is one of the hottest topics among economists these days. A sampling of some recent academic work on the topic:

Tim Groseclose and Jeff Milyo estimate how left-wing or right-wing media outlets are based on what research by think tanks they mention in their stories. They then compare that to the think tank research that elected officials cite when they talk on the House or Senate floor to calibrate where the media fits relative to the Congress. They find some interesting answers: most of the media does have a liberal bias (throwing out the editorial page, Wall Street Journal is the most liberal of all, even beating the New York Times!). Fox News is one of the few outlets that is right of center.

My colleagues Matt Gentzkow and Jesse Shapiro measure media bias by finding phrases that Democratic politicians use to describe an issue (“Estate tax” or “War in Iraq”) versus Republicans (“Death Tax” or “War on Terror”), and seeing which ones newspapers use. After building their index, they turn to determining what forces drive the newspapers’ choices. The answer seems to be that the newspapers slant their reporting to what their readership wants to hear, but they see little evidence that who actually owns the newspaper (e.g. Rupert Murdoch) matters for how it reports.

Finally, Alan Gerber, Dean Karlan, and Daniel Bergan run a field experiment to determine whether the content in a newspaper affects what people believe. They give free newspaper subscriptions to people in Northern Virginia. Some get the Washington Post (liberal) others get the Washington Times (conservative). They then poll the people in the experiment to see what their opinions are on political questions. Their results suggest that receiving the newspaper affects how people respond to the survey on a number of issues, especially who they want to be Governor.

Glenn Hauman

Giving newspaper subscriptions away leads to another interesting question that cues my media ecologist training.

When the cost barrier to entry disappears, as with, say, political blogs on the Internet, does there seem to be a stronger tilt to liberal reading? Put another way, is the New York Post gaining political mindshare because it sells for 25 cents when the New York Daily News and New York Times costs 50 cents? And how much of that translates to the web?


The last item is, indeed, the scariest. Given the nature of our education, political and media systems - it appears that people in those sectors are more likely to have a liberal bias. I'm not saying that that is wrong or right; however, it does provide three industries that have incredible influence on others. So the question is, "Can this ever be a system that is fair and balanced politically?" Sound like it's not something that will change any time soon.


This strikes me as one of those issues that will interest many but change very few minds. How one chooses to measure "liberalness" or "conservativeness" pretty well determines what the analysis will show.

I was struck by one of the examples: "Death Tax" v. "Estate Tax" If you take use of those two terms to denote "conservative" and "liberal" you ignore the fact that one is not only a political phrase, but a legal and technical term.

Kyle Gilman

Can you really claim liberal bias if a reporter uses a descriptive and accurate phrase like "estate tax" instead of a phrase created for persuasive political purposes like "death tax?" Or is it true what Stephen Colbert tells us that reality has a well known liberal bias?


Could media bias have little to do with conservatives and liberals? Why not investigate who is reporting the experiences of actual people versus who reports merely the political opinions of high government officers? Why not investigate the reporting of truth versus lies?

Three years ago, on a "left-wing" public radio station, someone who was actually in Iraq (not imbedded with military)-- someone who actually spoke with Iraqis-- told me that Iraq was in a "civil war". He reported that he saw no evidence of any significant number of the foreign fighters that Dick Cheney was describing as the main problem. I remember him because I was hearing little in the way of first hand reports. Last week, the Wall Street Journal finally used the term "civil war". Now, widely reported in newspapers are various definitions of "civil war", and based on several of these academic definitions, Iraq has been in a "civil war" since 2003. So, the little reported man in Iraq was accurate.

US big media reporting is an echo chamber. Both truth and lies bounce inside the chamber. Why lies get widely reported is to me a far more interesting question than the reporting of liberal versus conservative doctrine.



My colleagues Matt Gentzkow and Jesse Shapiro measure media bias by finding phrases that Democratic politicians use to describe an issue (“Estate tax” or “War in Iraq”)

I'm not an American or a resident of the USA, so perhaps I shouldn't be considered qualified to comment on this, but isn't the phrase "War in Iraq" just an accurate description of the situation rather than a political statement?

I'd point at Kofi Annan's recent warning of the dangers of (civil) war in Iraq as well as the actual, well, fact that people are killing each other there, but it seems almost disingenuous to do so...


Apologies. First sentence of my previous post should have been quoted.


As many have pointed out the use of slogans vs reality does not have anything to do with liberal vs conservative. Generally I've found the press to try and settle on the most neutral label they can find for a given group or issue.

The most obvious counter-example is the use of terms in the debate over abortion -- generally the media uses the preferred term of both sides, ie "pro-choice" and "pro-life", to describe them. They are both inaccurate and biased terms, but there is no succinct yet accurate way to describe both positions, so we just accept the preferred terms of both sides while rejecting the deliberately inflammatory labels they give each other.

Comparing such a situation to a specific legal term such as "Estate tax" and calling it "liberal" is absurd. Just as it is absurd to try and say that the War in Iraq, while it may be a part of a larger war on terror, can not be referred to specifically on its own. If anything, I would infer that the researchers themselves were biased if these are the best examples they can come up with of biased terms.



I don't know that they are judging the liberal/conservative slant of the actual terms, so much as measuring the media's use and repeating of the terms as used by liberals and conservatives.

Debating which end of the spectrum a word belongs on is not the point.

The point is measuring how often the vocabulary used by Mr. Liberal is used in the media vs. how often the vocabulary used by Mr. Conservative is used.


Off Topic:


Do you watch The Wire on HBO? I would love to hear your take on it.


I'd agree that even with the sometimes deafening noise from the right, the media has a liberal bias. (And rightfully so, as good journalism requires more of an open mind than many people on the right would appear to possess...)

However I think it is more important to recognize that the media has a bias. This seems unavoidable when media depends on advertising for revenue and the result is that we get stories which are sought after, written, and placed to appeal to the lowest common denominator of viewers/readers. I think this is why we get so many simplistic phrases & slogans, both liberal and conservative.


oops.... the end of the first sentence in the second paragraph if my previous post should read 'corporate bias'.


It's been a while since I looked at the Groseclose paper, but I recall having concerns about how he arrived at a definition of where the political center was (from which all other conclusions derive), as well as with the assumptions regarding the infuences of what was going on in the country over the time of the sample (I think that the time samples varied by media outlet, with some overlapping into the Clinton years and some only in the Bush years).

Such studies are interesting, but I think they do more harm than good if the methodology is not completely sound. Even a small shift in the definition of the political center in such a study can affect the perceived distance to the left or the right of the individual outlets. And since such studies generally regard bias as independent of events in reality, making no attempt at objective judgments of the actual validity of the positions, I think they will continue to remain flawed.



Most of those metrics are bunk, as others have pointed out.

The problem with assessing "bias" is that these things tend to assume that there are two equally valid sides to any issue, and the only reason a reporter would slant to one more than the other is because of some alleged bias on his part. The problem is that the two positions are rarely equal and sometimes there's just a right answer., by it's very existence, puts lie to the "liberal media" myth. But even that doesn't prove the media is conservative.

I've only ever seen evidence for three kinds of bias any news organization:
1) A sensationalism bias, designed to get viewers to watch (Missing white girls and celebrity trials fall here).
2) A bias towards advertisers, steering away from anything that would piss them off
3) A bias matching that of their audience.

In short, it's a "business bias", and politics doesn't enter into much. Even in the case of Fox News, it makes more sense to interpret them as going after a niche in a crowded market than anything else.

The big issue with the news isn't that it's biased, it's that it is a business, with the single goal of returning a profit to shareholders, rather than inform the public.



Debating which end of the spectrum a word belongs on is not the point.

On the contrary, that is the underlying assumption of the entire research! Just because the left or right uses a particular phrase more frequently does not make the use of that phrase indicative of bias. If the phrase itself is neutral, or the most succinctly descriptive available, I would expect it to be used by any reporter regardless of political affiliation.

If there exist only two phrases, and one is neutral while the other is biased, it is inaccurate to conclude that anyone using the neutral phrase is somehow biased *against* the beliefs of the proponents of the biased phrase.

Yet that seems to be the underlying basis of this entire study -- that every phrase represents an opposing end of the political spectrum -- rather than acknowledging some are simply descriptive or factual while others are deliberately pejorative.

A perfect example of this is the Fox News use of "homicide bomber" rather than "suicide bomber" to describe someone who delivers a bomb on his person. "Homicide bomber" is less specific (since it could apply to anyone who places a bomb intended to kill others). The use of the term "suicide bomber" is not indicative of a newspaper being liberal, it's indicative of them wanting to be informative.

Likewise, describing military action in Iraq as "The War in Iraq" is not liberal, it is descriptive. If you want to discuss global issues encompassing the entire economic and military struggle against extremists groups, then the "War on Terror" might be the most accurate term to use. But considering them to be interchangeable terms whose only difference is the bias of the reporter is both factually and ideologically inaccurate. Any study that starts with an assumption to the contrary strikes me as being biased before it even begins gathering data.


Tom Dunstan

Others have pointed out the problems in labelling 'War in Iraq' as a liberal phrase.

I read the paper which sampled think-tank citations. It was quite interesting, and the authors went to some lengths to remove possible systematic errors in either direction. The use of language was sloppy, though, with the abstract trumpeting 'bias', then part-way through pointing out that their definition isn't what most conservatives would mean when complaining about 'liberal bias' and suggesting that people use the word 'slant' to refer to deliberate bias, but then using 'slant' repeatedly in their discussions towards the end!

While it was interesting, and I would encourage people to read it, I have a number of reservations about their conclusions.

Firstly, there are a number of ways in which a news source might lean in a particular direction without revealing bias by citing think tanks. A news source which asserts a fact without bothering to back it up with a think tank citation appears more centrist than one which puts in the citation. Also, a news source might apply its bias by simply not running a story at all. An example of the latter is the Murdock-owned The Australian, whose editor apparently told at least one journalist that global warming stories simply would not get a run. Now, even if the paper had run the stories, it might have quoted organisations on both sides of the issue, so not running them needn't have pushed the paper's stats as calculated by the Groseclose and Milyo, but it's bias, nonetheless.

There might be lots of news stories which don't have an appropriate think tank to quote, and these are missed out entirely from the analysis. How much time a news source might devote to e.g. the Lewinski scandal or Mark Foley's recent scandal versus other breaking news might reveal bias in one direction or another, but would not show up in the paper.

There is interesting discussion about whether they have picked the middle of the road viewpoint properly (assuming thet their rankings are correct), but it is brushed off far too quickly IMO, given its importance to the paper's conclusion. The attempt to rank think tanks by their scholarliness vs activism doesn't seem to me a particularly useful metric, and the only other justification that they offer is that the three closest-to-center (as they had calculated the center) news sources provided moderators for the 2004 presidential debates. Unconvincing.

The final and probably most interesting point (to me) is that the authors are unable to account for the fact that most news organisations are centrist relative to the politicians who cite them. That is, most lay between the average Republican and average Democrat on the calculaed ADA scale. There should have been more news sources matching the supposedly extreme views of the politicians, if the news sources were generating their content for a given voters agreeing with those politicians. The obvious point to make is that that implies that the politicians aren't terribly representative of the populace. While the paper was clear to point out that it was trying to scale the news media against the average voter, the news sources produce news for a large portion of the population who do not vote. Recently (though perhaps not this year) there has been a tendency in American politics for the left and right to vote heavily, with the middle all but disappearing. That would explain the centrist papers relative to the politicians. The paper really has no data on the non-voting population, so it can't (and doesn't) make comment on them. Yet if that populace leans slightly to the left of the politicians that get elected (e.g. if, as has been suggested, the Republicans were better at "getting out the vote", at least 1994-2004), then that might match up with the supposedly left-leaning media, who suddenly dont look quite so left leaning as the paper supposes.



fairness both sides of the story to be told thats what most of us want.


I was suprised that the Wall Street Journal would have a liberal slant to its news. But quotes from think tanks are only one small part of the overall news. The news reporting is supposed to be unbiased, and in my experience, most news stories are. I think it is possible for a person with political views to write a story that is basically unbiased by simply sticking to the important facts without putting any spin on them. If the facts themselves happen to support a particular view, then that is not really evidence of bias. But then there is an issue if different people cannot agree on which facts are important and which are not.


Huslage: I think that's a pretty good rule of thumb, except as you point out, it's not scientific. As you imply, just because half the people believe it, doesn't mean that both sides are equally correct. I don't want to get into an intelligent design vs. evolution debate, but no matter how many people pile onto the ID side, evolution will always be more correct.

John Fembup

14 - "The big issue with the news isn't that it's biased, it's that it is a business, with the single goal of returning a profit to shareholders, rather than inform the public."

I would add that the news business routinely holds up the first amendment as their shield when anyone complains about their product. That constitutional shield and their use of it, make the news business unique among businesses.

I would also suggest that your description of business goals is based on a false dichotomy.

A business must offer a product that people will buy. Businesses cannot choose between making a profit and offering a product. A successful business must do both.

To say that the single goal of a business is to return a profit is to say e.g., that the Yankees show up at the Stadium, night after night just to keep score, and millions of fans pay to watch them keep score. Of course, that's nonsense.

The Yankees play baseball. People buy baseball. If the Yankees are good, they will make a profit. But first they have to play baseball. They do not choose between playing baseball and making a profit. Their success as a business requires both.

In the case of news media, the product is news. The reporting may be accurate or inaccurate, it may be unbiased or biased, but people expect news - that is what they buy. So a news business – or any business- just like the Yankees, does not choose between making a profit and serving a market. It must do both if it is to be successful.