If data are incorrect or incomplete, misclassification of groups and spurious associations may be introduced. Misclassification may affect both groups and masks the true association (random or non-differential). Misclassification may affect only one group leading to either over-estimation or under-estimation of an association (non-random or differential).