THE EFFICIENCY OF STATISTICAL TOOLS AND A CRITERION FOR THE REJECTION OF OUTLYING OBSERVATIONS