Reflections on writing about Confirmation bias

A year ago, the Wikipedia article on Confirmation bias was in a poor state. Whoever had written it was well-intentioned but they’d been working from a small number of sources and perhaps hadn’t seen the big picture. I started a substantial rewrite. The community gave me a lot of help to make the text accessible, and a couple of weeks ago it reached the highest quality standard on Wikipedia: Featured Article. (“Confirmation bias” as it was on 10th August 2009 vs “Confirmation bias” now).

This week I learned it has been chosen as “Today’s Featured Article” for tomorrow (Friday 23rd July). A one-paragraph summary will appear on the front page, where it can be seen by around four million users. Around sixty thousand will click through to the article itself. It will also be seen through the dozen or so sites that mirror Wikipedia. With this new prominence, it is more likely the article will be translated into other languages (extracts have already been translated into Spanish and Catalan). The are also other delivery platforms: I’m already planning a spoken version of the article, but won’t have time to do it before tomorrow. Being naturally the first Google hit for “confirmation bias”, the article has a high prominence (getting nearly a thousand hits per day) and it is regularly recommended and discussed on blogs, online communities such as Reddit.

So, it’s fascinating to watch the ripple effect of this article to which I’ve contributed. Confirmation bias is something you’d definitely hear about if you do certain courses within a psychology degree, but it’s not exactly the sort of topic that you would expect to see stories about in the newspaper or the evening news. Hence it’s significant that perhaps millions of people will hear about it through this article.  To be honest, this provokes mixed feelings.

Why confirmation bias in particular? In my final term as an undergraduate, I read Stuart Sutherland’s book Irrationality which introduced me to the research on cognitive biases and started the intellectual journey that led to my present interests. Two effects stood out for me. One was biased assimilation of ambiguous evidence, in which people with opposing views each interpret the evidence as supporting their points of view. The other was the closely related effect of illusory correlation. This is where people see associations between two things (such as homosexuality and gender confusion) that are not real but just reflect the person’s prejudice. These effects were the most worrying of all the bias research in that book, because they undermined the idea that biases (and the conflicts that result) can be lessened by showing people more evidence.

The lesson that some lay people take from bias research is just that “people are stupid”. My reading of confirmation bias is different: people are very clever in how they defend belief systems or stereotypes that they are somehow attached to. What’s more, we have an introspective blindness to how we are defending our beliefs: we don’t see what we are doing.

In my PhD, I read Klayman and Ha’s critique of the original confirmation bias experiments, which is one of the best academic papers I’ve ever read. It didn’t undermine the effects I’ve mentioned above, but exploded the idea that people just seek to confirm their hypotheses or beliefs. Much more recently, I read the chapter on confirmation bias in Rudiger Pohl’s textbook Cognitive Illusions, which clarified the whole area and suggested how I could structure a whole article. Raymond Nickerson’s review paper also gave a broad overview of how confirmation bias manifests itself. Detailed references are in the article.

I’m not expecting a greater awareness of confirmation bias to actually make people less biased. Already someone has tried to insert into the article a sentence about how people “are brainwashed to believe in the Holocaust”. We’ve also had the suggestion that the scientific theory of Anthropogenic Global Warming be used as an example of confirmation bias, as though it were the scientific bodies that we had to be suspicious of, rather than the AGW-denying bloggers and pundits. I expect that people with entrenched opinions will accuse other people of confirmation bias. It will just be another much like the politically sophisticated subjects in the Taber and Lodge study mentioned in the article.

So even though making people aware of confirmation bias might not reduce its incidence, I still think it’s a worthwhile thing to teach people about and an interesting thing to learn about. For one thing, it might lead some people to take an interest in rational skepticism. Ever since Francis Bacon condemned the biased evidence for omens and superstitions, the topic of confirmation bias has been illustrated with ideas from the paranormal. I hope the article prompts people to think about the basis for claims of the paranormal, which is mentioned tangentially in several points of the article. Just because someone is sure they have seen it, doesn’t mean that it’s true or that there is good evidence for it.

What I hope people will take from the article is something that is shown rather than stated, namely the power of the experimental method in psychology. Before I studied psychology, I thought that it would be unproductive to try to study the human mind with science, but that was just my lack of knowledge and lack of appreciation of the psychologists’ very clever experiments. Another subtle lesson is the strength of the cognitive approach which treats human beings as processors of information. This steers a course between a rational-choice approach and a more biological, reductionist approach.

I’m enthused to write more about cognitive psychology for Wikipedia, and hopefully my writing style is less obtuse and inaccessible than before I took this plunge.

  1. #1 by michaelwebster on July 23, 2010 - 2:49 pm

    I briefly looked at the article, and I think it is very good.

    I have two questions, and one observation.

    1. The section on Wason breezes over the inability of people to apply the logic of if then correctly. This could be expanded. Why did you focus more on the three number experiment?

    2. Why is there not section devoted to cognitive dissonance and confirmation bias? One and same phenomena? Or what distinctions should we draw?


    1. The section on illusionary associations would be better if you explained with numbers why there is a slight negative correlation between arthritis and rain. Go through the counting, point out that in this world there is a lot of arthritis and a lot of rain, but that when it rains 66% of people report arthritis, while in the non raining world, 75% of people report arthritis.

    This explanation would also help people understand that in evaluating a conditional, if .. then, you need to look at all four truth functional possibilities also.

    Very nice article!

  2. #2 by Martin Poulter on July 26, 2010 - 12:24 pm

    Thanks for the constructive suggestions, Michael. There is a bit of an overlap with cognitive dissonance, but they are separate topics. Reduction of cognitive dissonance is a drive, while confirmation bias is a group of effects. Sometimes the drive can be responsible for the effects, as in the political MRI study. Cognitive dissonance is potentially a much broader topic than confirmation bias, though.
    I didn’t want to spend a lot of time on the historical stuff as that might be confusing in an article that’s mostly about the present state of knowledge. Arguably the 2-4-6 task is a cleaner demonstration of the original point as it doesn’t involve the question of how people interpret if-then statements. I’d like to see a long and complete article on the Wason selection task eventually, but it’s on a long, long “to-do” list.
    Your suggestion for the section on illusory correlation is very good. I’ll take a look at implementing it.

  3. #3 by michaelwebster on July 27, 2010 - 1:05 am

    1. I see your point with cognitive dissonance – surely one of the greatest drives we know of.

    2. For the purposes of your article, I think you are right about the 2-4-6 task as opposed to the if-then. A full article on Wason would be very useful.

    3. Look forward to the changes on illusory correlation, and I will make future suggestions on wikipedia.

  1. Deliberate wrongness: a strategy to avoid confirmation bias « Bias and Belief
  2. How to write a Psychology article for Wikipedia « Bias and Belief

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: