Information bias occurs when there are systematic differences in the way data on exposure or outcome are obtained from the various study groups. This results in inaccurate data collection, leading to errors in the estimation of the association between exposure and outcome.