Confirmation Bias (On human Cognition)

Two of the most common cognitive mistakes, Confirmation Bias and Attribution Error, have far reaching consequences in everything from politics to auto repair.

Attribution Error is where, given an absence of contradictory information, people will assign positive motives to themselves (or people like them), and negative motives to others (or dissimilar people) for any given act.

The most common example is driving. When we cut someone off, we say we did it accidentally, because we were in a rush, or, in some cases, because they deserved it.

However, when someone cuts us off, we say they did it because they're a jerk or a bad person. This is opposed to equally probable explanations such as they did it accidentally, because they were in a rush, or because, in some cases, we deserved it.

This is why there are no guilty people in prison. Convicts usually claim that while other prisoners may deserve to be punished, their own crimes (even murder) were justified.

Any "immoral" act (stealing, accepting bribes, lying under oath) by a political figure will be seen by his party in the most sympathetic light possible, as an aberration, mistake, or lapse of judgment. The opposing party will see the same offense as a sign of failed moral character, and a symptom of rampant corruption. That is until a member of that party commits the same act.

Confirmation Bias is the human tendency to only try to prove theory, not disprove it. This is differs from the tendency to try to avoid being proven wrong, as confirmation bias applies even in the genuine search for new knowledge, not just to the testing of established beliefs.

Imagine 4 cards, each with a number on one side, and a letter on the other side:

E    7    4    F

If someone (lets call her "Susan") says that "Every card which has a vowel on one side has an even number on the other," which cards would you have to flip over to see if Susan is right?

Think about it before moving on???

The way to determine if Susan is right is to see if every card with a vowel on one side has an even number on the other. If even one vowel has an odd number, then Susan is wrong.

"E" is the most common first card flipped, and indeed is correct. If the number on the other side of "E" is odd, then Susan is wrong. If it is even, that doesn't mean Susan is right- just that she right so far.

"4" is the second most common card flipped. This is wrong, and is our first encounter with confirmation bias. What is on the other side of "4" is irrelevant. If it's a consonant, that doesn't make Susan wrong (She didn't say ONLY vowels had an even number on the other side). If it's a vowel, that's consistent with Susan's statement, but doesn't prove it, or give us any information we don't already have.

"F" is, by the same token, irrelevant.

The "7" card is not flipped by most people. This is another mistake. If the other side has a consonant, it means nothing. But if it's a vowel, then Susan is decidedly wrong. This card must be flipped to determine if Susan is right, but people ignore it because of confirmation bias.

When al-Qa'ida successfully destroyed the world trade center, they saw it as evidence that God was on their side. When they failed to simultaneously blow up 6 airliners over the Atlantic, they didn't consider it proof that He wasn't. That is confirmation bias.

Rev. Jerry Farwell claimed that God allowed the 9/11 attacks to occur because of gays and abortion rights. Florida was later hit with the worst series of hurricanes in its history, with Miami-Dade getting disproportionately high damage. Despite this coming on the heels of serious irregularities in Florida's 2000 election, mostly centered on Miami-Dade, he did not claim that God allowed it as a result of voter fraud. That is attribution error.

One political example where attribution error and confirmation bias combine and confound is in claims of political bias in the media.

Each side of the political spectrum believes that the media is biased against them. The Pew Charitable Trusts Project for Independence in Journalism found that in the 2004 election John Kerry got a higher volume favorable press than George Bush. Most right wing media organizations have cried loud and oft that this is proof of liberal bias in the media.

To assume that favorable press is an indication of bias is an attribution error. Favorable press could just as easily come from a candidate's charm, his ability to cover up indiscretion, genuine moral superiority, or time in the public eye (4 years in office gave greater opportunity to dig up dirt on Bush), or any number of things. The right simply assumes it must be because the media is biased.

Assuming for a moment that favorability did indicate bias, there is yet another problem with using the 2004 Pew study to show liberal bias. That problem is the 2000 Pew study, which found that in the 2000 election the media gave Bush nearly 50% more favorable coverage than it gave Gore. Pew called the studies mirror images of each other.

The 2004 study references the 2000 study, and they are on the same website. Yet The Washington Times, Fox News, and other right wing news sources are continually claming that the 2004 study proves a liberal bias. Given the 2000 study, the claim borders on ludicrous.

Are you liberal? Are you reading this right now? Enjoying it? The right wing sucks, don't they? Knowingly lying about what the 2004 study means??? Of course, to say that is to suffer attribution error, when it is far more likely that the right's oversight is simply confirmation bias.

Original