Media bias and the predictive model

Like many people with views in the right half of the American political spectrum, I have given some thought to the question of media bias. It does seem to exist. Most major media outlets are considerably to the left of the political center-of-gravity of the American electorate, and this significantly influences their coverage of events.

I'm not much interested in arguing this point in this post. For a pretty solid recent piece of social science research on the subject, you might read this. (Betsy Newmark comments cogently here.)

What interests me here is how this bias actually functions. Journalists, of course, publically place great value on even-handedness, and I believe they are mostly sincere in this. Yet they apparently do not practice this virtue in their journalism. How does this happen?

One factor, of course, is that the community of reporters is pretty uniformly liberal. This means that most of the people with whom a reporter discusses his work have biases that confirm his own. In such an "echo chamber", attitudes and approaches that from the outside seem slanted and unfair can pass muster as sensible and balanced.

We mentally locate assertions on a continuum from "undeniable fact" to "unsupported opinion", and we deal with them accordingly. What I am suggesting is that people who live and work among like-minded colleagues will naturally and unconsciously shift statements they like toward the "fact" end of the scale, and statements they dislike toward the "opinion" end of the scale. So if everyone around you agrees that "Bush is a moron," then this sentiment becomes more fact-like for you and so can easily creep in as a presupposition of your news article, despite your abstract commitment to journalistic fairness. (A similar effect, even stronger, can be seen in the faculties of institutions of higher learning.)

Nota bene -- I am not saying that everything is just someone's opinion, or that all opinions are equally arbitrary. Some things are facts. Some judgments based on the facts are more reasonable than others. The "Bush is a moron" meme strikes me as poorly supported by the available evidence. It seems rather to be a sort of playground taunt, like saying "You are fat and ugly" when what you really mean is "I dislike you and the things you do." The rhetorical purpose is to deny the enemy anything that might be considered a virtue. Nevertheless, there are otherwise smart people who take Bush's moronhood as a solid fact and use it as a way of understanding political events.

Lots of people think that unconscious bias in the news media functions in this way. I have a somewhat different theory. It is based on two observations. First, the practice of journalism involves making choices -- what story to cover, what facts to include, etc. Second, although journalists may (like anyone) wish to see their political side prevail, they are probably (like anyone) even more motivated by the desire for success. And what brings success to a reporter or an editor or a news organization? Journalistic success: providing the definitive coverage of an important event or development. But that is trickier than it sounds.

Look at it this way. You are an editor for a major news organization. On any given day there are maybe a hundred things that you might do a news story about, but you can only give prominent coverage to a dozen of them. You have to decide which of these events is likely to be most important. What criterion do you use? Well, the events do not occur in isolation. Each of them is a development in a larger story-line. Some of these larger story-lines will just peter out; others will prove to be crucial turning points in history. You'd like to identify early which story-lines will be most significant, and concentrate on them.

Similarly, within a given story, there are hundreds of facts that might be reported. You'd like to include the facts that will be most important in determining how the story turns out.

But of course, all of these decisions about what stories to tell and how to tell them must be made before things have turned out. You can't wait until next year to report on this year's economic news. You have to decide now how much attention to give to, say, rising fuel prices. If the economy is really strong next year, everyone will say, "Well, those fuel prices weren't such a big deal." But if the economy slides toward recession, everyone will say, "Fuel prices were a critical factor." Both would be reasonable valuations after the fact. But, dagnabbit, as an editor you have to make the "no big deal" versus "critical factor" choice today, before anyone actually knows how things will go. So you make an informed guess and go with it.

(There are other factors at work, of course. Saying that high fuel prices threaten to destroy the economy may be the more exciting option, regardless of the probabilities. Bad news sells newspapers and is, let's face it, more fun to tell. But most journalists are serious people who do not cynically slant their stories just to boost circulation, and to hell with the facts. I don't think they do that, anyway. Or not very much.)

Here is my thesis: the judgment that journalists and editors must make when choosing and assembling stories is essentially a prediction of the future. These people have a model in their heads about how things are likely to happen, and this influences how they cover present events.

Consider Iraq. News stories out of Iraq over the last couple of years have been pretty gloomy. On the other hand, most members of the US military who are actually in Iraq, or who have come back from Iraq, are optimistic about their mission. The news media sees a deteriorating situation. There are bombs all the time, and hundreds of people are killed each year. The political situation is fractious and teetering on the edge of civil war. US forces are showing the strain, over 2000 killed so far. Folks in the military reply something like this: The violence is mostly confined to a small part of the country. There are new local governments in place and three successful national elections in a year. There are new schools, new infrastructure, a tremendous revival of the Iraqi economy, etc. The bad guys have shifted their attacks from 'hard' targest (like US Marines, who shoot back) to 'soft' targets (like Iraqi schoolteachers, who don't) -- a definite sign of weakness, and a shift that is costing them dearly in popular support. The net result is that soldiers and marines are often astounded by the pessimistic cast of the coverage in the US media.

Of course, the real situation is complicated. Good and bad things are happening. Car bombs kill innocents, and new schools open at the same time. In a given week, suppose three car bombs go off and ten new schools open. The news media goes with the car bombs on page 1 and largely ignores the schools, because the reporters and editors generally believe that the bombs are more likely to be decisive to the long-term outcome of the Iraq war than the schools are. In other words, the media has guessed the end of the story already, and they are doing their best in their reporting to lay the groundwork for that ending, to better inform their readers.

If news judgments are made on the basis of a predictive model, an implicit forecast about the future course of events, then present events that do not conform to the model will appear to be "noise" and will to some extent be filtered out. This is done for the noble purpose of helping the reader "make sense" of events, but what it means in effect is that the news is automatically shaped to support the model. So the whole enterprise depends on the accuracy of the a priori model.

Where does the model come from? Some people think that it is just created out of personal biases, pure and simple. The predictive model is spun out of what the journalists want to happen, or fear will happen. I think that this view does disservice to the journalists. By and large, I think they try to base their predictive models on facts. And there's the rub. Because only a small fraction of the facts a reporter has comes from his own direct observation. The rest mostly comes from other media reports -- and these are shaped by the other predictive models of the other journalists.

Therefore, we would expect an arbitrage process at work among the models used by journalists who read one another. Reporter X believes that the economy will decline next year, and uses this judgment to select facts for a story about the weakening economy. This story is read by other journalists. They are more aware of the facts that X emphasizes and less aware of the facts that X filters out as "noise". This influences their own views. Pretty soon, the predictive models have a high level of agreement. Conventional wisdom is born -- not because anyone is consciously grinding an ideological axe, but because they have all developed the same conclusions from the reported facts. But of course, those facts have been selected for reportage precisely because they support a cluster of guesses about the future.

Ideology plays its role, of course, since it influences what future scenarios one finds plausible. The predictive models that the news media are using are not just facts or deductions from fact. And this is the place where the left-of-center politics of the journalists themselves can affect their coverage, in spite of their efforts to be fair-minded. They are not hypocrites with their thumbs on the scale, substituting partisan propoganda for unbiased reporting. They are just trying to help their readers understand and anticipate the world. They are as surprised as anybody when it doesn't work out the way they thought it would. But by that time, of course, it is no longer news.

The answer? Like most interesting problems, we can't fix it; we can only manage it better. The media should be more aware of, and skeptical of, its own set of predictive models. Journalists should say to themselves every morning, "I do not know how all this is going to come out." Because they don't, really. And they should be held accountable for their models. If a reporter spent early 2003 breathlessly reporting how the invasion of Iraq was going to lead to tens of thousands of US casualties in the first six months and a refugee crisis involving millions, then this should not be forgotten. (This is one of the best things that the blogosphere does -- and one of the things that the mainstream media finds most disconcerting.)

I want to conclude by mentioning two places where I think that the media are in fact a bit more culpable for bias. In the first place, I think that reporters have a less-than-defensible desire to fit the facts into a small number of pre-existing story templates. For instance, there are any number of reporters in Washington who are on the lookout for the Scandal That Brings Down Bush. After all, this is what happens to two-term presidents in the modern era. A huge scandal and cover-up in the second term always destroys or diminishes an administration: Nixon and Watergate, Reagan and Iran-Contra, Clinton and Monica. It's a standard (and, for the news folks, highly profitable) format. So, like Californians feeling the start of a tremor, they look at each new story, however trivial, and wonder, "Is this the Big One?" The result is a lot of foolish hype of non-issues and, over time, the vague impression that the present administration is particularly scandal-ridden -- when, by historical standards, it is not.

Second, I think reporters are somewhat dishonest about what they are doing. On the one hand, they say that they want to "make a difference in the world" through their reporting. On the other hand, they also say that they stand apart and observe -- that their job is to cover events, not to influence them. The media loves its influence but cannot frankly admit to it, because its moral standing (such as it is) relies on being "outside the fray". Such doublethink can only promote other sorts of dishonesty and subterfuge, which opens another door to partisan bias.

This can get silly. A New York Times reporter recently stirred up controversy while working on a report about on-line child pornography. (Betsy Newmark notes, and comments, here.) The reporter met and spoke with a kid who was trapped in this world by predatory adults and a drug habit. The reporter went so far as to put the kid in touch with the authorities, which eventually led to the arrest of some pretty scummy people. The controversy was, of course, that the reporter had sacrificed his holy objectivity by intervening in the events he was covering. My naive reaction was, the better man he. And journalists are human beings (and, yes, citizens) before they are journalists.