Rankings Redux

z-score-with-cumluative-percentiles-that-match-book.jpg

Since I posted the last entry, I have had an interchange with Newsweek that has cleared up some of the mystery. The order of the rankings in the Green Corporations table is determined by something called a z-score, not the raw impact data. The z-score is a measure of the distance a score is from the mean, normalized by the standard deviation.* It’s just something like this that may have triggered Lord Disraeli’s initial outburst about statistics. It is more work that I am willing to do to calculate these scores for the whole set, so I will just have to accept the mathematical accuracy of the data. But not of the significance of the data.

Mother Nature, as I noted, cares only about the upset we are causing her, not about the intentions or reputation a company possesses. She might laugh out loud at seeing the data on which this ranking is based reported to two decimal places when everything is based on opinions one way or another. Everything that has gone into this ranking is arbitrary from the use of revenue to normalize impact scores to the choice of weighting factors and the use of z-scores. It is ingenuous to conceal this arbitrariness behind four significant figures (for all the nerds that read this blog).

Anyway, I do recant my statements about the internal consistency of the presentation, but not about its relevance or meaningfulness.

*Z-scores are used in rankings, such as the US News and World Report’s ranking of law schools and business schools. It, in theory, shows how any entry compares to the average for the factor. The composite scores are very sensitive to the weighting factors used. Why, in this case, the choice of 45%, 45%, and 10%? Nice round numbers???

|