Biased selection of case and controls in a study, and difference in phrasing questions could lead to a difference in the response of case and controls, resulting in spurious association. Also this artificial association could result from errors of study design e.g. misclassification of study or disease, measurement errors, loss to follow-up and observer bias.