What Is Information Bias?

Information bias (measurement bias) is systematic error in how exposure, outcome or other variables are measured, leading to misclassification and distorted results. It occurs even when the right people are included.

Types and examples

Differential versus non-differential

If misclassification is similar in compared groups (non-differential), it often biases results toward no effect; if it differs between groups (differential), results can be biased in either direction.

Reducing it

• Use validated instruments and objective measures.
• Standardise procedures and train staff.
• Blind assessors and, where possible, participants.
• Use records made before the outcome occurred.
• Pilot-test questionnaires.

Illustrative example

Invented for teaching. In a case-control study of a rare disease, cases search their memory for past exposures more thoroughly than controls, so exposure appears more common among cases. Using pharmacy records for exposure would reduce this recall bias.

Misclassification in routine records

Studies using hospital records or administrative codes depend on how accurately diagnoses and procedures were recorded. Codes may reflect billing practice rather than clinical truth. If you use such data, check whether the codes have been validated, sample some records to estimate accuracy, and describe the likely effect of errors in your limitations.

Common pitfalls

• Relying on self-report for sensitive exposures.
• Not reporting validation of the measurement.

Key takeaways

• Information bias arises in measurement; blinding and objective measures reduce it.
• Think about direction as well as presence.

This page is general educational information about research methods. It is not medical, legal or regulatory advice; follow your institution's ethics board and your target journal's instructions.