Overview

Once we believe something, we can be quite good at finding reasons to keep believing it.

Confirmation bias describes the tendency to seek or interpret information in ways that support what we already believe or expect. It can also influence what we remember. Evidence that agrees with our existing view can feel more convincing, while contradictory evidence may receive greater scrutiny, be explained away or simply attract less attention. [1]

Imagine you've decided that a particular app is difficult to use. You notice every moment of hesitation, every misplaced tap and every frustrated expression. When someone completes a task without difficulty, you might attribute their success to experience or luck. Both observations are real, but they're being interpreted through different expectations.

Confirmation bias doesn't necessarily involve conscious dishonesty or an unwillingness to consider alternatives. People can sincerely believe they're evaluating information fairly while applying different standards to evidence that supports or challenges their views. Studies of how people assess conflicting evidence illustrate this tendency, although the findings don't mean everyone responds the same way in every situation. [2]

The effect matters wherever people form judgements from incomplete or ambiguous information. In design, research and product development, those judgements can influence which problems are investigated, whose experiences are considered representative and which solutions are pursued.

In practice

Confirmation bias can appear throughout the research and design process, sometimes in surprisingly ordinary ways.

A researcher who expects users to struggle with a new navigation pattern might pay particular attention to hesitation during usability testing. A product team convinced that a feature will improve engagement might interpret positive feedback as validation while dismissing complaints as unusual cases.

The problem isn't necessarily the observations themselves. It's the way expectations influence which observations receive attention and how much weight they're given.

Research practices can make biased interpretations easier to notice and challenge. Teams can document assumptions before testing, agree on how they'll evaluate findings, use neutral questions and look carefully at results that contradict their expectations. None of these steps guarantees an unbiased conclusion.

It's also useful to distinguish between an observation and its interpretation. A participant pausing before selecting an option is observable behaviour. Concluding that the participant was confused is an interpretation, and one that may need further investigation.

Considerations

Confirmation bias is sometimes presented as a problem that can be solved by becoming more objective. Unfortunately, recognising a bias doesn't automatically make someone immune to it.

Nor does every instance of maintaining an existing belief demonstrate confirmation bias. Some beliefs are well supported, and not every piece of contradictory information deserves equal weight. Even looking for evidence that fits a hypothesis can be a useful way to test it, depending on the circumstances. [3] The important question is whether the evidence is being evaluated fairly and whether the approach could reveal that the belief is mistaken.

Attempts to counter confirmation bias can introduce problems of their own. Deliberately seeking disagreement is useful, but treating all opposing perspectives as equally credible can create false balance. The quality, relevance and strength of evidence still matter.

In research, the goal is to create conditions in which unexpected findings can genuinely change our understanding. That requires more than asking people to set their assumptions aside. It means making those assumptions visible and using methods that allow them to be challenged.