Print Shortlink

Confirmation Bias In Action: The BBC turns 10% into 1%

I was going to let this pass without comment. In my experience, biased as my experience probably is, I find a lot of journalism involves somebody telling us what we should think, as justified by selective, edited, manipulated and misinterpreted data. Never is this worse than when politics is supposedly bolstered by science. And when I use the word, I really mean ‘science’. There is science in this world, but some ‘science’ gets the name despite a scarcity of repeatable experiments and an inability to reach consensus about what would count as evidence for, or against, any theory. I could not let it pass, so let me tell you a story about confirmation bias, politics, journalism and ‘science’.

One reason for my difficulty, despite my desire to let it pass, was that only last week I wrote about confirmation bias and the press, noting how economist and columnist Paul Krugman was buffoonishly unaware of the possibility that he was suffering from a severe case of confirmation bias when he wrote a list of reasons explaining why he does not suffer from confirmation bias. It is a scientific fact that confirmation bias is ubiquitous. Though this really is a scientific fact, it is important to recognize that most people, including journalists, continue to pretend, or believe, that they do not suffer from this bias. Krugman’s example is extreme, but when it comes to politics and journalism, there appears to be no widespread moral prohibition against journalists who present biased arguments, in order to please their biased audiences. On the contrary, that is the essence of the business model for most purveyors of news and comment.

In many respects, we should find it appalling to live in a world of incessant, unrepentant bias, especially when the bias extends to misusing the scientific method and misinterpreting data in order to reach a pre-determined conclusion. However, in a world of many voices, we can let it pass. We can reasonably assume that the many opposing biases will cancel themselves out, like waves that fitfully and energetically clash but leave nothing permanent behind. If readers do not like the New York Times, they might read the New York Post, though if they want to counter bias they should probably read both.

However, not all voices present themselves as one amongst many. Some present themselves as uniquely superior to all. This is why the BBC is really the worst source of news in the whole world. With its mid 20th century understanding of human psychology, the BBC continues to pretend it can be an impartial public service broadcaster. It tells us it never suffers from bias, in anything it produces, and if it does, the bias was only a momentary slip. The BBC believes it has found a way to cure every single one of its 23,000 staff – the largest workforce for any broadcaster in the world – of the pernicious evil of confirmation bias. And when challenged as to how they have accomplished this herculean task, they reel off the reasons, with the kind of blissfully ignorant irony that makes you wonder if they will employ Paul Krugman when readers of the New York Times finally tire of him.

So as I mentioned above, I was going to let this pass, but confirmation bias is important, and underreported. It is hence unlike most stories that feature in the news, which tend to be relatively unimportant, and grossly overreported. Confirmation bias clouds our judgement, leads to bad decisions, makes the world a worse place. Democracies will never have good public policy unless our societies tackle and minimize confirmation bias. But our institutions let the public down, and the BBC is the worst example of that collective failure. And this failure grows more acute when bad science is used to justify poor conclusions. Many people realize the world is grubby and full of narrow-minded and prejudiced people. They look to science as a less polluted source of information. It is therefore reckless to abuse the scientific method, and to undermine future trust in science by misrepresenting its conclusions.

Whilst the story I am about to tell is a small story in many ways, it is an important story about bias, and the failings of the institutions that are supposed to support our democracy. I could let it pass when some bad scientific interpretation caught the eye of a journalist due to his confirmation bias. But I would not let it pass when his biased story caught the eye of many members of the public due to their confirmation bias. And so, because of this chain of events, a scientific ‘truth’ is now known to a large number of people, confirming what they already thought they knew about the world, even though the science does not, in fact, tell them what they want to hear. And for this little tragedy, this small defeat in the war for truth, we should thank the ‘impartial’ BBC. So please indulge me, as I now dissect what is today’s most shared story on the BBC News website, a piece entitled “Study: US is an oligarchy, not a democracy“.

This is how the story begins.

A review of the best commentary on and around the world…

Today’s must-read

The US is dominated by a rich and powerful elite.

So concludes a recent study by Princeton University Prof Martin Gilens and Northwestern University Prof Benjamin I Page.

This is not news, you say.

Perhaps, but the two professors have conducted exhaustive research to try to present data-driven support for this conclusion.

The offending journalist was correct in one regard. This is not news. However, it is not news because the ‘data-driven support for this conclusion’ is flawed in one important respect: the data. A lot of you, by now, will already be suffering the onset of confirmation bias, so I ask you to retain a cool, objective, scientific outlook as I continue.

The academic paper in question presents no data about any ‘rich and powerful elite’. I know this, because I read the paper. If the journalist read the paper, then he deliberately omitted to mention the most important caveat in the paper. Instead of reviewing data about a ‘rich and powerful elite’, it presents data about the top 10% of earners in the United States of America, and shows a correlation between their views and the policies adopted by government. This is very clear throughout the paper, but its importance is stated in this excerpt.

We believe that the preferences of “affluent” Americans at the 90th income percentile can usefully be taken as proxies for the opinions of wealthy or very-high-income Americans, and can be used to test the central predictions of Economic Elite theories. To be sure, people at the 90th income percentile are neither very rich nor very elite; in 2012 dollars, Gilens’ “affluent” respondents received only about $146,000 in annual household income. To the extent that their policy preferences differ from those of average-income citizens, however, we would argue that there are likely to be similar but bigger differences between average-income citizens and the truly wealthy.

There is no need to analyse the rest of the paper, because everything hinges on this assumption. The researchers indisputably show a correlation between the policies implemented by the American government, and the opinions of the top 10% of Americans by income. That is a fact. There does appear to be a relationship between opinions expressed in surveys, the income of the person expressing the opinion, and the likelihood that the US Government will adopt policy that conforms to that opinion. The imbalance is clear – the top 10% get what they want, the rest do not. But the authors have to assume that the goals of an ill-defined ‘rich and powerful elite’ are the same as 10% of the entire population. And they also have to assume that if the government adopts a policy it is only because of the influence of this ‘rich and powerful elite’, and not the influence exercised by 10% of the entire of population.

Is evidence of an imbalance that favours 10% of America’s population equivalent to the writer’s meaning, when he reports that the ‘US is dominated by a rich and powerful elite’? No. Even the academics distinguish between the ‘truly wealthy’ and the top 10%. And here, on this single, simple point about data, the paper’s conclusions must fall. I suspect the weakness of the professors’ arguments was in their mind, when they used weasel words in the conclusion, also repeated by the BBC’s journalist:

Americans do enjoy many features central to democratic governance, such as regular elections, freedom of speech and association and a widespread (if still contested) franchise. But we believe that if policymaking is dominated by powerful business organisations and a small number of affluent Americans, then America’s claims to being a democratic society are seriously threatened.

In other words, Americans do seem to live in a democracy, but if policymaking is influenced in a way not proven by the data shown in the paper, then the claim to be a democracy would be threatened. Notice how we leapfrog from some data about the top 10% of American earners, to a conditional statement about when democracy might be vaguely ‘threatened’, to a BBC headline which says the USA is not a democracy. The BBC’s hack follows this misleading headline with a complacent comment, saying that the USA’s status as an oligarchy is so widely appreciated, it is not even newsworthy. What is the role of the BBC in this case? Is it to open our minds to new information? Seemingly not. Is it to encourage some people to close their minds, by telling them what they want to hear? I suspect so.

You could reasonably argue that this academic paper draws a conclusion based on data. You could reasonably argue that this academic paper states a conclusion about the influence of the truly wealthy. But you cannot reasonably argue that the paper’s conclusion about the influence of the truly wealthy is based on the data it presents. To argue otherwise is evidence of one of two faults: foolishness, or bias. It is perfectly possible to imagine an oligarchy where 10% of the population dominate. That would look a lot like apartheid South Africa. In that country, the whites were richer than blacks, and held power. But we also understand there were significant differences in wealth, and opinion, amongst the whites. That is not the kind of example this journalist has in mind, when asserting the following about the USA.

…the wealthy few move policy, while the average American has little power.

I think we all know the real reasons why words like ‘few’ and ‘average’ are being presented to the reader, when it would have been just as easy to write about ‘ten percent’ and the other ‘ninety percent’. I would not think of ten percent of the population as a ‘few’ people. The research demonstrates an imbalance, but not the kind of imbalance that the professors wanted to show, and thus their own bias is revealed. On the contrary, the top ten percent of the American population will include a lot people like journalists and professors – the kinds of people who might tell us that their power and influence has been exaggerated, and that all power resides with an elite that excludes them.

It is telling that the British Broadcasting Corporation finds it necessary to highlight a minor academic story from the USA at the same time as the British Labour Party employs David Axelrod’s services as a political communicator. Axelrod is famous for helping Barack Obama to win two Presidential elections. News coverage of Axelrod’s appointment has uncritically repeated Axelrod’s message about his appointment. Axelrod says he is not some mercenary for hire. No, no, no. He really shared all the views of all the candidates he worked for, and was paid handsomely by. Yes, yes, yes. He shared all their views, even when he swapped between opposing candidates. Yes, yes, yes. And now he shares all the views of the British Labour Party, and is not working for them because he is cynically motivated by the prospect of a big pay day. Phew. Because nobody in the BBC wants to suggest that Labour’s campaign message of ‘us’ versus ‘them’ is going to be orchestrated by a guy who makes so much money from politics, he definitely counts as one of ‘them’.

Axelrod’s message, uncritically repeated by the BBC, is that Labour will win by saying similar things to Obama. But as ever, the political memory is weak, which may also be a sign of confirmation bias. It is not possible for Labour to simply emulate the message written by Axelrod, and spoken by Obama, during his two successful Presidential campaigns. That is because two different messages were given. Readers of Halfthoughts from its inception may recall that its banner has subtly changed over the years. Before Obama’s first victory, it depicted President George W. Bush, with an empty thought bubble, to represent his empty head. When Obama won in 2008, I thought I would play a trick on his over-adoring fans. I changed the banner to make it appear I was uncritically repeating the central message of his 2008 campaign: change. It was a slow joke, that grew steadily funnier over the years. By 2012, when little had changed, I delivered the utterly predictable punchline, crossing out the word ‘change’. Romney’s defeat has saved me the effort of needing to alter the banner since, and so the joke rolls on.

When it came time for the American public to contemplate change, it was predictably time for Axelrod to change his message. Change and hope were no longer in fashion. It was time for ‘better the devil you know’. But devils are not very popular, just like Obama was not very popular, as measured by his very low approval ratings. So Axelrod needed to present Obama as a combatant against a foe who was even more unpopular. The campaign message became ‘us’ versus ‘them’, where ‘us’ was understood to be very many, and ‘them’ was understood to be very few. This, we now learn, is also the message Axelrod will use with Ed Miliband. And this makes perfect sense, when we consider that the polls say Ed Miliband is even less popular than Barack Obama.

When analysing the ‘us’ and ‘them’, there is always some deliberate vagueness about who is included in each group. This is not properly compensated for by the intensity of the emotion used to confirm that ‘we’ are wonderful honest hard-working victims, whilst ‘they’ are miserable evil selfish persecutors. When Axelrod and Obama were campaigning in 2012, some said the few were as many as 1% of the population. No Democrat dared to suggest that ‘they’ were 10%, although Romney made an imbecilic error by opposing himself to 47% of the population. If ‘they’ were 10%, that would raise awkward questions about who donated to Obama’s campaign. If ‘they’ were 10%, they would include very many journalists, businessmen and celebrities who campaigned vigorously for Obama. The 10% clearly includes people like Paul Krugman, who writes a column subtitled ‘The Conscience of a Liberal’ but who is very definitely near the top end of the 10%, if not in the top 1%. The game of ‘us’ versus ‘them’ only works if we deny the possibility that some of ‘us’ are really ‘them’.

And so, Axelrod’s message for Labour will be the same as the one he used for Obama in 2012, presumably because Labour is unable to tell us much about what they would change. Like Obama’s presidency, it will be important to elect them because of what they represent, and who they oppose, but not because of the policies they espouse, and what they will do. Suggesting you might actually do something will inevitably lead to disappointment, and that difficult shift of message that asserts the time for change has passed, and it is better to stick with the devil you know.

Amidst this fracas, the BBC plays its dutifully ‘impartial’ role, by highlighting some research, out of all the thousands of political science papers published all the time, that just happens to accord with Axelrod’s message that he fought a rich elite, in order to preserve America’s democracy. Do I see confirmation bias here? Yes I do. Perhaps I perceive my own confirmation bias. But like Krugman, I doubt that my own confirmation bias is the fault here. I do not blame myself for confirmation bias, because I can do something done by neither Axelrod, nor the BBC, nor the professors who wrote this paper. I can offer a second, alternative model that equally fits the data, but which definitely does not fit their interests.

Who would be in the 10%, either in the US or UK? Axelrod, definitely. Most senior management at the BBC, definitely. The better-paid BBC journalists, probably. US professors of political science, very likely. Every American senator, representative and governor, whether Republican or Democrat, with absolute certainty. The entire front bench of the Labour Party, beyond doubt. Every Member of Parliament is in the top 5% of UK income, just because of their basic MP’s salary, before counting all the top-ups and perks they may qualify for. So when these researchers observed a correlation between the opinions of the top 10% of the US, and the decisions reached by government, I find it easy to propose a simple model which would explain this correlation, without needing to assume the influence of a super wealthy elite.

Everybody in the American government, of whatever political persuasion, is in the top 10%. They have more in common with other people in the top 10% than they do with anyone else. Most of their friends are in the top 10%. Their parents were probably in the top 10%. And their views are bolstered by professionals, businessmen, academics and journalists who are also in the top 10%. Not surprisingly, they share the views of most people in the top 10%, and they make policy decisions accordingly. That is the character of American politics. That is the character of British politics also, although today’s Labour Party sometimes pretends otherwise.

Journalists and professors are turning a blind eye to a perfectly plausible explanation of this important data. The data shows there is a link between an American’s income, and the decisions made by America’s government. They ignore the most straightforward explanation, because some want to believe a corrupt elite holds sway over all of us, and none in a position of power or influence wants to admit what the data most obviously suggests. All of these people belong to the 10% class, and the data implies their class governs and controls everything. I admit my alternative model is no more proven by the data than the professors’ model of a tiny dominant elite. But at least I am open to thinking it, and stating it as a possibility. What excuse do they have, for failing to admit the possibility that governments inherently favour the interests of the kind of people who actually run governments? What excuse could they have, apart from bias?

Leave a Reply

*