I loved this one:
"All true students of stat know that the mean is pablum, while the MEDIAN is the red meat for the adult statistical palate."
Both the mean and the median can be misleading due to particular characteristics of the set of values to which they apply, but the arithmetic mean is the standard summary function for sets of values because the mean is dependent on the actual values within the set. The median is affected not by the values themselves, but by only the ordinal position of the values in the set when ordered. Values (outliers) that are radically different from the majority of values in the set may skew the value of the mean from the apparent center of the distribution of values, but the scope of the set can be adjusted to eliminate outliers if desired. The median value can also be very misleading if a set of values includes many members that are near the maximum or minimum value of the set and there is a large difference between the maximum and minimum values.
For example, if the Notre Dame offense in five games in a month scores 3 points, 3 points, 41 points, 41 points, 41 points, the median is 41, but the average is 26. Another team scores 3 points, 3 points, 7 points, 62 points, and 63 points and their median will be 7 and their mean will be 28. Using the median makes Notre Dame look 34 points better in analysis than the other team when the other team actually outscored the Irish in total points. In this case, the mean is more accurate and "mean"-ingful than the median.
I noticed that in the table of offensive output for the months of September, October, and November, Duranko uses the median value for the months of September and November, but his total yards value for October is actually the mean for total yards in that month. The median value for total yards in October is 396. Gotta keep the apples with the apples and the oranges with the oranges.