How are confounding and bias defined, and what methods are used to control each?

• A. Confounding is a random error; bias is a systematic error; control by increasing sample size.
• B. Confounding is a mixing of effects by a third variable; control via randomization or stratification. Bias is systematic error in data collection/selection; control via blinding, standardized protocols.
• C. Confounding is mixing by measurement error; bias arises from sampling error; both controlled by blinding.
• D. Confounding and bias are the same; control by adjusting for confounders.

Answer: (A)

Confounding happens when a third variable is related to both the exposure and the outcome, so the observed association is distorted by that extra influence rather than by the exposure itself. To control confounding, use design or analysis methods that balance or adjust for the third variable: randomizing participants so the confounder is evenly distributed across groups, stratifying the analysis by levels of the confounder, restricting the study to a limited range of the confounder, or adjusting for the confounder statistically in multivariable models.

Bias is a systematic error in how data are collected, measured, or selected, which pushes results away from the truth. Controlling bias involves reducing systematic differences: blinding those involved in data collection or assessment, using standardized and validated measurement protocols, training assessors consistently, ensuring complete and unbiased follow-up, and pre-specifying analyses to avoid data-driven decisions. Increasing sample size lowers random (sampling) error but does not eliminate bias or confounding.