Confirmation Bias: Why Human's Don't Change Their Minds

Everyone has had a moment where they've disagreed with someone and spent hours finding proof on the internet that their side of the argument is the right one. You may come across tens of articles that prove you wrong but you just ignore those and find the ones that agree with you. There is a name for this tendency and it shows up everywhere.

What is Confirmation Bias?

In psychology, confirmation bias is defined as seeking or interpreting evidence with a bias toward an existing hypothesis. In simpler words, humans tend not to change their minds even when presented with conflicting evidence. Confirmation bias refers to the subconscious cherry-picking of data. This sets it apart from, for example, a debater who consciously builds a case in support of an argument. The leading theory behind why our brains do this points to two effects: it saves energy by sparing us from comparing competing ideas, and it prevents cognitive dissonance (the discomfort our brains feel when we hold, or are confronted with, conflicting beliefs or information).

Where Do We See Confirmation Bias?

In the age of social media, confirmation bias can be reinforced by the way algorithms customise our feeds. If you click on or engage with a post, the algorithm feeds you similar content, so people often see media that favours their position in an argument. This can make users believe their position must be the 'right' one, as it seems to be the most popular. The result is a modern version of confirmation bias.

There are two kinds of confirmation bias: motivated and unmotivated. Motivated confirmation bias occurs when we want a particular conclusion to be true and favour evidence that supports it. Unmotivated confirmation bias occurs even without a preferred outcome, simply because our existing beliefs shape how we search for and interpret information. For instance, when asked to test an idea, most people would naturally only try to find cases where it is true rather than also comparing cases where it is false.

How Does Science Uncover Confirmation Bias?

There are a number of ways in which scientists can call out confirmation bias in each other's research. Two main methods are p-curve analysis and replication.

p-curve analysis: In scientific research, results are generally regarded as significant if the p-value is less than 0.05. The p-value is the probability of getting results at least as extreme as those observed if there were no real effect. When a hypothesis is true and a study is well run, the significant p-values across many studies tend to bunch up at very small values, well below 0.05. However, if researchers are biased toward one conclusion, they may (often unintentionally) tweak their analysis, for example by excluding certain data points or trying several statistical tests, until their p-value falls just below 0.05. This is called p-hacking. When many such results are graphed, there is an unusual pile-up of p-values just under 0.05 instead of an even or tapering spread.

Replication: If research is unbiased and sound, fellow researchers should generally be able to repeat the study and reach the same result. If the original data or analysis was biased, the findings often will not replicate, although a failed replication can also have other causes, such as chance or differences in method.

Confirmation bias is a characteristic of the human brain that appears everywhere, from everyday discussions and politics to science. It is important to constantly question and challenge the things we learn. The ideas that survive those challenges are what advance ideas in our worlds.