Bias: Combatting confirmation bias

Purpose: To provide an understanding of confirmation bias and ways to avoid it, in order to support good decision making on complex problems.

Description: Confirmation bias manifests as follows: 

  1. We accept a story uncritically if it confirms what we would like to be true and reject it if it contradicts what we would like to be true.
  2. We accept a fact as data, even if it is an outlier which is not backed up by large-scale data.
  3. We accept data as evidence, even if it is consistent with rival theories, without looking for evidence that supports one theory and rules out all rival theories.

In brief:
A story is not fact because it may not be true.
A fact is not data because it may not be representative (especially if it is only one data point).
Data is not evidence – you need to check if it is also consistent with rival theories.
Evidence may not be proof because it may not be universal.

Three tips to overcome confirmation bias, to ensure you have reliable evidence:

  1. Seek other viewpoints, especially from people you disagree with. Create a culture that encourages dissent. Encourage ‘devil’s advocates.’
  2. Find experts who have gathered evidence, with a focus on high quality evidence. Review their credentials and check where their work has been published, as guides to quality.
  3. Pause before sharing. Ask yourself: if the opposite finding was published by the same experts in the same journal, would you still be willing to believe it and share it?

Reference: “What to trust in a ‘post-truth’ world” by Alex Edmans was a talk at ‘TEDxLondonBusinessSchool’, London, UK in 2017. Video (18 minutes) online at either:

  1., or
  2. YouTube:

Related tools on the i2Insights blog:

Related tools on Wikipedia:

Posted: February 2019
Last modified: March 2019