The Law of Regression

"Hiding in plain sight," says Daniel Kahneman. According to the author of Thinking, Fast and Slow, regression effects are so common that we do not even see them. Like the air we breath, regression is invisible. I hope to show why leaders need to work harder to make the invisible visible.

As you may remember, regression to the mean is the statistical phenomenon that explains why children whose parents are extraordinarily tall tend to be shorter. For any set of quantifiable characteristics, there is an arithmetic average of the observed characteristic. Successive measures of the observed characteristic tend to approximate the average, not the extreme. This is the Law of Regression.

Why is increased awareness of the regression effect important to leaders generally and school leaders in particular? There are several good reasons, but I will confine my thoughts to one.

We know that human performance is a characteristic subject to observation and measurement. And we know that we've spent a king's ransom on instruments specifically designed to measure observed performances, for example, North Carolina End-of-Course and End-of-Grade Tests for students.

Normally-matriculated students take a particular EOG or EOC test only once, yet from this sample, the public makes inferences about the student population as a whole and the impact of teachers on those students from one year to the next. In fact, we have included progress on these tests in the evaluation of teachers and principals. Whether fair or not, you may infer for yourself aided by this text.

Here's the thing: The Law of Regression cares not that different students took the tests. Nor does it care whom or what got better or worse. The Law of Regression does one thing. It aggregates the blob of bodies to whom the tests are administered, year after year, and with steely logic demands that variation around some floating average results in a mean to which all successive scores will tend.

Someone check me on my own logic, but is it not the case that, given the Law of Regression, we might err when we attribute improved scores to improved teaching and identify the former as causing the latter? Equally, is it not the case that we may err when we attribute a decline in test scores to something gone horribly wrong at school when, in fact, it is simply random variation? These are the stories we tell ourselves because we must make sense of things. Honestly, we must. Luck is for losers, right?

Undoubtedly, the Law of Regression applies to organizations and organizational results far beyond public education. Rather than reflect on the Law of Regression when it favors us, however, we leaders pat ourselves on the back for our contribution to building a better employee. Of course, if performance declines, it was the employee's fault. Isn't it grand to be human?