Innumeracy

If you have not read Innumeracy by John Allen Paulos, but down the internet and go to your local library or bookstore now. The internet will be here when you get back.

I read a copy from my library. I'm going to buy a copy to keep and re-read, and to share with my kids when they get a little older.

I do a lot of personnel management. One section of chapter 4 really struck a chord with me, and I think will make me a better manager. It will also certainly effect my plans for Democracy 2.0.

Paulos applies the probability theory of 'regression to norm' to human behavior. Regression to norm indicates that if there is some baseline average result (like an average number of units sold) and at any interval there is a high deviation from that average (like a particularly good or bad sales month), probably the following interval will see results that are closer to the average. For instance, if you are averaging 30 units sold per month, and you sell 60 this month, you are highly likely to sell less than 60 next month. That is the definition of "average", after all.

You can substitute trend for average, and the same thing occurs. If you normally sell 1 unit more each week than you did the previous week, and suddenly you hit a spike and sell 5 units more this week, then you will likely sell less than 5 additional units next month.

And it works backwards, so if you sell less than your trend one period, you will probably sell closer to your trend the following period. Once you know the average or trend, you can plot that probably your results will be closer to the trend after any large deviation.

It doesn't matter what the activity is. It doesn't matter what the underlying causes are, as long as there are no real structural changes that effect the environment. If you have an average or trend that you can track, then it will behave this way.

Now apply that to people; specifically to performance. If you are managing a person who averages X units over a certain period, and they have a markedly up or down period, they will probably perform something closer to their average the next period. If that weren't true then the "average" you thought you were looking at really wasn't, or there was some environmental change that you didn't recognize.

But you are saying, "what if they just got better?"

Learning curve

Standard deviation

Trend line

Praise and Punishment

"behavior is most likely to improve after punishment and to deteriorate after reward. Consequently, the human condition is such that ... one is most often rewarded for punishing others, and most often punished for rewarding them." researchers Tversky & Kahneman

What to do? Recognize the trends and standard deviations. Manage for behavior changes that carry the person up the learning curve--change the trend.