After reading the article "The Flaw of Averages," I agree with certain statements made by the author, but I am not in complete agreement with the methodology the author used. In the article, the author does present a number of valid examples to support his opinions, however I would counter that there are a number of issues that could have been considered that were not.
The author tries to make a case that averages should not be used when you "need a number." I think using an average when you "need a number" is fine as long as it is used appropriately. Two criteria I would use in determining appropriateness is
1) Averages should be clearly and accurately defined. The more this criteria is used the more accurate the "average" number will be. An effort should be made to specifically identify what the average represents.
2) Data should not be random - that is there is some correlation as to the quantity of items being measured to something else.
For example - one can expect the sale of lawn mowers to increase during the months of March, April, May & June and decrease in November, December, January & February.
If these two criteria were considered in the examples in the article, the author may have decided not to use them after all.
Consider the statistician that drowns. This example leads one to believe he was in a deeper part of the river or near the center. Using the information provided, we just do not know. Had the statistican considered my criteria, the example would have been stated differently. Although the river is on average three feet deep, rivers generally start shallow at the banks and get deeper the closer one gets to the center. This is a correlation of the...