Selection bias occurs when there are systematic differences between the comparison groups in a study. For example, differences in clinical signs between the groups might lead to different disease progression or response to treatment between groups, rather than the intervention itself. Proper randomisation and/or blinding have not been achieved, which can affect the statistical analysis and internal validity of the study.
Selection bias is sometimes used to refer to an error in the selection of studies for reviews. Publication bias is a type of selection bias. Confusingly, selection bias is also sometimes used to refer to systematic differences between the study group and the general population. This leads to problems with external validity.