Confirmation bias and how to avoid it

Part 1: Everyone has a bias

Rhiannon Davies | September 15, 2021

This is the first of a two-part series on how to keep confirmation bias from affecting your writing. In this post we’ll talk about what confirmation is and the different types of biases you can fall prey to. Part 2 will talk about how confirmation can affect your writing, and what you can do to avoid it.

In a real-life example, I can tell you that my cat Monster is scientifically proven to deter tigers. I read it on the internet, and all my friends agree with me, so it must be true.

If I cuddle Monster once a day, I won’t get eaten by tigers. I read an article that said that tigers are very sensitive to the smell of domestic cats, and will avoid them at all costs. The tigers will smell her on my clothes and run away. The article’s right — I haven’t even seen a tiger outside a zoo, let alone been attacked by one. And it’s all down to my brave little Monster.

Of course, the fact that Aotearoa New Zealand doesn’t actually have tigers roaming freely is the real reason why I haven’t been eaten by tigers. The article I mentioned earlier is made up. But it’s a good example of confirmation bias.

What’s confirmation bias?

Think of a subject where you have a strong opinion or position. You will be more willing to accept information that reinforces your opinion than to accept information that goes against it. That’s confirmation bias.

Everyone has confirmation biases — they’re part of the unconscious heuristic process that we use to make decisions about the world and how we work within it. Confirmation biases show up in the way people look for, interpret, and remember information.

Different types of confirmation bias

Confirmation bias comes in a few shapes and sizes:

• biased search for information
• biased interpretation of information
• biased recall of information.

Biased search for information

Biased searches for information are where someone searches specifically for information that reinforces their opinion, and ignores all other information. It doesn’t matter if the information they find is reputable or not.

Both Monster and I agree that cuddling her once a day (at least) is the best thing I can do with my time. If someone thinks I spend too much time cuddling her and not enough time doing other things, I’ll want to find something that proves that cuddling her is a better use of my time than the other things.

A quick Google search for ‘scientific benefits of cuddling cats’ yields 21 pages of hits for articles about why cuddling cats is scientifically proven to be good for you. Most of the hits are from lifestyle magazine-style websites whose information you can take with a grain of salt, but some of the hits are from reputable websites like Healthline. These sites have a lot of sound, believable information on them. Sounds like Monster and I are right about cuddling being the best thing to do with my time, right?

Performing a search using the same keywords on Google Scholar gives a completely different result. Google Scholar only stores information on peer-reviewed academic articles and books. Research on Google Scholar has been rigorously tested using the scientific method before publication.

Read about the scientific method on Wikipedia

Only two of the results that Google Scholar pulls up on the front page are about the benefits of living with cats. The 100 pages of results are mostly veterinary texts about different illnesses and other health problems cats suffer. Disappointingly, little real evidence appears to support my claim that cuddling Monster is the best way I can spend my time.

Fortunately for me and Monster, a Google Scholar search for ‘scientific drawbacks of cuddling cats’ has no relevant hits whatsoever. Nothing in published research shows that cuddling her is a worse way to spend my time than doing anything else.

Biased interpretation of information

Biased interpretation of information is where people take the meaning that is more closely aligned with their opinion from information. The meaning they take might not be the full meaning of the information.

Let’s say someone publishes an article about tortoiseshell cats having ‘tortitude’ — a very strong will and feisty temperament compared to cats of other colours. The article talks about what tortitude is and has a list of characteristics. It doesn’t talk about whether tortitude is positive or negative: it just describes it in terms of what tortoiseshell cats do and how they behave.

I find tortitude endearing because of Monster. I like the fact that she’s sassy and takes no nonsense from anyone, including dogs bigger than her. I would find the article about tortitude that uses terms like ‘stubbornness’, ‘independence’, and ‘unpredictability’ a positive one.

Someone who prefers a more chilled-out cat would not find tortitude a positive character trait. If they read the same article, they would feel that a tortoiseshell might not be the cat for them. They might take the article as a warning.

The article’s information hasn’t changed between me reading it and the other person reading it, but how we interpret the information is completely different.

Read the tortitude article at Taylor and Francis Online

Biased recall of information

Biased recall of information is where someone remembers only certain parts of information that reinforce their opinion on something, even if they find the information in a neutral way.

Let’s say I’m having a conversation about Monster with my colleagues at work, and one of them says ‘Monster is very cute, but my cat is cuter’.

When I’m recounting the conversation to my flatmate when I get home from work, I might only remember my colleague saying ‘Monster is very cute’. ‘Monster is very cute’ is the part of the conversation that aligns best with my world view, so it sticks in my head better.

Join us next month to find out how confirmation bias affects your writing — and how to avoid it.

Learn more about biases

Take our Critical Thinking workshop
Read our blog post about critical thinking, bias, and media manipulation