Data bias

Last updated Jun 5, 2023

A type of error that systematically skews results in a certain direction.

Related:

• Fairness

Types

• Sampling bias
• Observer bias (experimenter bias/ research bias)
  • The tendency for different people to observe things differently
• Interpretation bias
  • The tendency to always interpret ambiguous situations in a positive or negative way
• Confirmation bias
  • The tendency to search for or interpret information in a way that confirms pre-existing beliefs

Type of bias:

• Sampling bias is when a sample is not representative of the population as a whole. For example, maybe your sample did not include people above the age of 65. Or maybe you excluded people from certain socioeconomic groups.
• Observer bias is the tendency for different people to observe things differently. For example, stakeholders from different parts of the world might view the same data differently and draw different conclusions from it.
• Interpretation bias is the tendency to always interpret situations that don’t have obvious answers in a strictly positive or negative way, when, in fact there is more than one way to understand the data. Data that does not provide an obvious set of conclusions makes some people feel anxious, which can lead to interpretation bias. For example, a team member might interpret inconclusive survey results negatively, while other team members might be able to think more carefully and assess the data from different angles.
• Confirmation bias is the tendency to search for or interpret information in a way that confirms pre-existing beliefs. For example, you might ask only specific stakeholders for feedback on parts of your project because you know they are the most likely to have the same perspective as you.