Experiment bias

Types of bias

A bias is a systematic error in sampling. There are numerous types of bias which can inadvertently influence the results of a statistical test:

  • Cognitive bias: Biases which may stem from emotional or moral motivations or from social influences which deviate from rationality in judgement
  • Confirmation bias: The tendency to search for and interpret information in a way which confirms your pre-existing beliefs or hypotheses
  • Implicit bias: The error which occurs when assumptions are made based on personal experiences and opinions that do not necessarily apply more generally
  • Observer bias: When a researcher subconsciously influences participants or results to match their expectations
  • Recall bias: When the respondent hasn’t remembered things correctly
  • Recency bias: When an event that has happened most recently is disproportionately over-represented in the results
  • Reporting bias: When the frequency of particular events, often unusual ones, are over-represented in the data because they are more readily reported
  • Selection bias: Accidentally working with a subset of your audience when you believe you have a representative sample
  • Sponsorship bias: The tendency of a scientific study to support the interests of the people or organisations funding the research
  • Survivorship bias: The error of concentrating an experiment on observations which made it past some selection process and not having visibility and therefore overlooking others that didn’t