September  2018

Everyone believes that media are biased, but rarely will people from opposite sides of any issue agree as to which way the bias lies. And it's even rarer for media to admit to any bias at all.

So, here's an objective and quantitative way to measure media bias:

But first, some background.
Bias can take different forms:
The most obvious is telling lies, as practised by some national propaganda organisations. Joseph Goebbels (Hitler's propaganda chief) is an historical example, while North Korea, Russia and China are contemporary practitioners. Less often, commercial media will tell out-and-out porkies, sometimes to support a cause but more often to boost readership. News of the World's(?) "B29 found on the Moon" - complete with picture- is a famous example of this.

Bomber on the moon
Short of outright lying is "spin"; in which the facts remain fairly intact, but interpretations can be wildly different. For example, depending on the political slant of the news source, white farmers are either the victims or the villains in the South African land expropriation issue.
More insidious is selection bias. By this, news providers don't tell lies and don't necessarily distort the facts. Rather, they present or give prominence only to those stories that favour their political preferences. Even reputable newspapers like the New York Times, the Washington Post and the UK's Guardian use this method to push their causes. Radio New Zealand are champions of selection bias; their panels are loaded with "progressives", and their interviewees have stories favouring a particular slant- usually victims of greedy businesses and brave activists standing against the power of America or Israel- or multi-national corporations. If they were slanted the opposite way their panels would be outraged by socialism, their heroes would be small business people done over by the bullying state, and their villains would be welfare bludgers.

Does this matter? Except in places like North Korea and truly undeveloped countries, people now have access to multiple news sources, and sufficient education to discount spin and identify lies. But it's much harder to counter selection bias, because when all or most of the media in a country slant one way, those who don't agree with the official line feel that everyone's against them- and get angry.

This situation now exists in many Western countries, where media tend to be to the Left of the populations they serve.

Ideological Placement of Each Source's Audience
It's thought that this has happened because when media (and academia) have even a few Left leaning people in senior positions, they tend to hire people whose views align with their own, until all the dissenters are gone- while Right wingers do not select in this way. This is called O'Sullivans law: And the Left believe their capture of the media is just as it should be, because the world will be a better place when everyone agrees with them, -which will happen with enough "leadership" (read censorship and intimidation).

But history suggests to me that this is a dangerous, because those in the wider population whose views do not align with the mainstream media, don't like having their views continually disparaged and suppressed. Their groundswell of anger underlies the election of Trump, Brexit, and the rapid growth of populist parties in the UK and Europe. The Left should therefore be concerned with media bias, even when it's in their favour, because it's sowing seeds of discontent that will likely become violent ( already has:

in which the anti-immigration reactionaries are to a significant extent expressing their frustration about media and leaders who "aren't listening".) Which is why I've been thinking of ways to measure media bias that even the most partisan will find difficult to deny or ignore. Not that this by itself will settle any of the issues of substance in our ever more divided communities, but if media can be persuaded to move back towards the centre, some causes of discontent will diminish: Those on the Left will be disabused of their belief that they are an overwhelming majority and those on the Right can start to feel that their views are getting something like equal time- and that they're not up against some vast conspiracy.

Politics is not a single dimension; there are divergent views on multiple issues disposed each side of a centre. Left wingers can be social conservatives, Right wingers can be social liberals (often are), and authoritarianism is a disease infecting both extremes, though currently more virulent on the Left.
This test can be used to measures the difference in views between a media organisation and average opinion along multiple axes.
It measures the extent to which individuals, governments, organisations and media, exploit the discomfort most people feel when 'everyone disagrees with them' to influence opinion. They do this by labelling opponents as more extreme (and therefore worthy of less support than they have) while representing those with favoured views as more centrist and moderate than they are -so as to claim more support for them than they have.
The test works by counting the labels that media apply to people they are reporting on, capturing their bias score by how much they inflate the extremism of those they oppose and moderate their descriptions of those whose causes they support. When looking at Left/Right bias for example, the score is composed of a Right bias number and a Left bias number. The ratio of these numbers is the bias score.
The Left bias number includes how many times people on the Right are described as 'extreme Right' or 'far Right' and etc and the Right bias number includes how many times those with Left wing views are described as 'far Left', 'Marxist" and so on.
Added to these basic scores are numbers from how the subjects of news stories are characterised. For example, Left wing media typically label Jordan Peterson as 'alt Right' and a "pop psychologist" - to denigrate and devalue his views (on the Right he's an "internet phenomenon" or "public intellectual" to boost his influence). Use of a derogatory label scores bias points.
And a third source is reporting of polls: News organisations tend to publish polls which support their cause: Fox quotes Rasmussen on Trump's approval rating because it's generally higher than Gallop, vice versa for The Huffington Post. Selection bias in poll reporting therefore also scores points.
For testing other biases (women's rights or religion for example), the labels counted will be different, but the method is the same.
Using an extreme label counts as a bias point even if the label is true, because generally only opposing media use such labels- supporting media use one that's more moderate (but less accurate). For example, an unreconstructed marxist is very likely to be labelled as such by the Right wing Info Wars site (which therefore scores a Right bias point for this), whereas Mother Jones (a Left wing site) would most likely soften this to 'socialist' or 'social democrat', so as not to scare the horses.
The numerical score from counting these things is, I believe, a good proxy for the direction and extent of bias exhibited by any news organisation.
For more objectivity (and less tedious counting), the entire process could be digital.
In addition to counting the frequency of use of extreme labels, a program could record every label used to describe known personalities, rank them from the laudatory to the pejorative along the axis of an issue, then use these for scoring - no judgements then being required as to where the subject's opinions lie.
And there'll be ways that I haven't thought of to make the scores even more robust.
Every radio program, every television program, every podcast every blog, every opinion piece and even every news release by governments, corporations and NGOs could be run through such a program, and a bias score published. It even works for news aggregators, because if they link more often to one side, the summed scores will measure this.

Nor will gaming a system such as this be easy, because labelling is at the very heart of how media with an agenda push their causes.
This, or something similar, could be a useful public service- especially for on-line blogs and podcasts.


PS: And here's a thought; if there was an updating register of media and commentator bias, it would be easy to also measure selection bias as a two number score - just count how many times Right wing sources and Left wing sources are linked or quoted.