Why Medians May Not be the Message – for Talent Data

I was reading a piece in the Journal this morning on a scientific debate regarding an Alzheimer’s study. The debate centers around the use of the median as a descriptive summary statistic. The article brought me to my own recollection of Nassim Taleb’s own objections to some types of quantitative analysis in The Black Swan, and then to an article Stephen Jay Gould wrote in the 80′s that I had never read, and felt ashamed I hadn’t since I was a huge fan of his.

The article was called, “The Median Isn’t the Message”, a nod to Marshall McLuhan, and is a personal story of his own diagnosis with a deadly cancer. The median survival rate of the cancer was 8 months… but being an evolutionary biologist, he knew to dig into the research properly to understand what was really going on. One quote:

But all evolutionary biologists know that variation itself is nature’s only irreducible essence. Variation is the hard reality, not a set of imperfect measures for a central tendency. Means and
medians are the abstractions. Therefore, I looked at the mesothelioma statistics quite differently – and not only because I am an optimist who tends to see the doughnut instead of the hole, but primarily because I know that variation itself is the reality. I had to place myself amidst the variation.

It’s that variation I worry about as we think about talent analytics. The essence of innovation, high performers, and more are not from a central tendency – they lie in the extremes.  I’m still forming my own outlook on this topic, and welcome everyone’s thoughts.

Here’s Stephen’s article in full : http://people.umass.edu/biep540w/pdf/Stephen%20Jay%20Gould.pdf

More Read

A Complete Guide to Overcoming Executives’ Concerns about Hadoop

Great Video on How Econometricians Think About Talent Data

Data Design Matters

SAP Spends Big on SuccessFactors for Cloud Computing and Talent Management

Three Critical Junctures