Confirmation bias

What it is

Confirmation bias is the tendency to seek out, interpret and remember evidence in ways that favor what we already believe or expect (Nickerson, 1998). It is not one single habit but a family of them: what we search for, how we read ambiguous facts, and which details stick.

An everyday example

Sam is convinced a new coworker, Jordan, is unreliable after one missed deadline. Over the next month, Sam notices every late reply from Jordan and mentions it to others. When Jordan finishes a project early, Sam assumes someone else must have helped.

Sam is not lying. Sam simply files each new event under the story that already exists.

Why it happens

Testing an idea by looking for supporting examples feels natural and takes little effort. Looking for ways it could be wrong takes more work and is less pleasant, especially when the belief matters to us.

Ambiguous evidence also leaves room for interpretation, and we fill that room with our expectations. Over time, small tilts in what we search for and how we read it can add up to great confidence built on a one-sided sample (Nickerson, 1998).

How to counter it

• Ask what would change your mind. Name, in advance, the evidence that would count against your view.
• Test, don’t just collect. Look for cases where your idea predicts something specific that could fail.
• Keep score fairly. If you are tracking someone’s behavior, record hits and misses in the same place.
• Invite a critic. Ask someone who disagrees to make the strongest case against you, and listen for its best point.

What the research shows

In Wason’s (1960) “2-4-6” task, people were told that the triple 2, 4, 6 fits a hidden rule and could test other triples to discover it. Many proposed only triples that fit their own guess, such as “numbers rising by two”, instead of triples that might break it. Of 29 participants, only six announced the correct rule (any increasing sequence) without first naming a wrong one.

Nickerson’s (1998) review gathered evidence for confirmation-type tendencies in many settings, from everyday social judgment to science, medicine and law. The review also notes that the term covers several different processes, so it is better treated as a broad pattern than a single, precisely measured effect.

Testing ideas by checking confirming cases is not always wrong. It becomes a problem when we stop there and never look for disconfirming ones.