In clinical studies, errors are usually categorized into systematic errors (bias, confounding) and random errors. Systematic errors are of particular concern because they lead to over- or under-estimation of effects. Bias is a systematic error that distorts the relationship between exposure and outcome. As a general rule, it is not possible to mathematically adjust for biases once they have been introduced. Confounding is a special case when the distorted relationship between cause and effect is caused by a third factor. Confounding can be neutralized using careful study design or statistical methods. Validity is the opposite of bias, and is usually divided into internal validity and external validity. A study is internally valid if the estimates drawn from the study population are free of confounding and bias. A study is externally valid if its results can be applied to a separate population. Internal validity is a prerequisite for external validity.