The answer to this question is: groups have equal and balanced composition.
Which means that each groups in experimental research are not displaying the too mcuh difference in term of average characteristics.
This will make sure that the data that acquired during the research are stable and can be trusted (not affected by a special variance)
