kənˈdɪʃən the procedure that is varied in order to estimate a variable's effect by comparison with a control condition