Recognising and Addressing Bias

Bias can affect how information is selected, produced, interpreted or presented. It may lead us to favour certain evidence or perspectives without adequate justification. In research, it can also arise from methods or reporting practices that systematically distort findings.
Having a particular perspective, theoretical approach or personal experience does not automatically invalidate a source. Examine whether its assumptions are clear, its evidence supports its claims and important limitations or contrary evidence have been overlooked.

Peer-reviewed research can still be affected by bias. Examine how the research was designed, conducted, analysed and reported.

Watch for:

• Bias in design or data collection: selection procedures, leading questions or measurement methods that systematically distort findings
• Selective reporting: presenting favourable findings while omitting relevant outcomes or analyses
• Publication bias: the available literature disproportionately representing particular results, such as positive or statistically significant findings

In quantitative research, a small sample can make an estimate less precise without necessarily making it biased. Judge whether the sample and methods are appropriate to the research question.

A particular theoretical framework is not itself evidence of bias. Consider whether its assumptions are explained and justified, and whether relevant alternatives are addressed.

Check funding and conflicts of interest, but examine their possible influence rather than treating them as proof that the findings are unreliable.

AI-generated answers can reflect or amplify biases in training data. An answer may sound neutral while presenting a limited or distorted view of a topic. It may:

• Give disproportionate attention to dominant cultural or social perspectives

• Reproduce stereotypes relating to gender, race, disability or other social groups

• Omit relevant evidence, alternative interpretations or the perspectives of people affected by an issue

When checking an AI answer, consider whose perspectives are represented, whose may be missing and whether important evidence or limitations have been overlooked. Compare its account with appropriate original sources, giving weight to the strength and relevance of the evidence.

If the output summarises or compares sources, evaluate those sources and check how accurately and fairly the AI has represented them. Agreement from another AI tool is not independent verification.

For the full checking process, see Evaluating AI outputs. For broader guidance on other risks, see AI Risks and Limitations.

Here are some examples of how bias might appear in different subject areas:

Legal Studies: Selecting only cases that support your preferred interpretation while overlooking relevant contrary authority can distort your analysis. Identify the applicable jurisdiction and explain which authorities are relevant to the legal question.

Creative Practice: Cultural or aesthetic biases may affect how design trends or artistic works are evaluated.

Applied Fields (e.g. Business, Computing): Be aware of industry bias — favouring particular vendors, tools, or methodologies due to commercial influence or dominant paradigms.

Diversify your sources: Seek relevant evidence and credible perspectives that challenge your initial view, including voices your sources have overlooked. Weigh them by the strength and relevance of their evidence. You do not need to give unsupported claims equal weight.

If sources disagree, see Dealing with conflicting information for help comparing their evidence and explaining the differences.

Use objective criteria: Evaluate evidence based on clear, consistent standards rather than personal preferences or pre-existing beliefs.

Engage in reflective practice: Regularly reflect on why you chose specific sources and consider how different choices might change your conclusions.

Seek feedback: Talk to peers, tutors, or writing support staff to uncover blind spots in your thinking. Others can help you spot assumptions you may have overlooked.

Ask yourself these questions when researching or writing:

• Have I sought evidence from multiple credible sources, even those challenging my initial views?
• How might my personal experiences or beliefs be influencing my interpretation?
• Are there credible arguments or evidence I might be unintentionally dismissing?

To build on your understanding of bias, explore these related guides:

• Spotting misinformation and disinformation: investigate potentially false or misleading claims
• Evaluate information: assess a source’s evidence, relevance and suitability for your task
• Evaluating AI outputs: check claims, reasoning and omissions in AI-generated material

This guide is part of our critical thinking and information evaluation series.