What does the ratio difference in means divided by the mean absolute deviation tell you about how much visual overlap there is between two distributions with similar variation?