The Stick is Mightier than the Carrot

Psychologists will tell you that carrots are a far more productive strategy than sticks.  You might think this is a blindingly obvious statement, but the facts don’t always support it.  Sometimes the psychologists are proven wrong.

That is exactly what happened to Daniel Kahneman.

In the 1960′s he was giving a lecture to a group of Israeli air force flight instructors about the psychology of training, showing how experiments with pigeons had demonstrated that rewarding good performance was far more effective than punishing poor performance.  His belief was that when it comes to learning and development the carrot is far more powerful than the stick.

He had hardly finished when one of the flight instructors blurted out

“With respect Sir, what you’re saying is literally for the birds…  I’ve screamed at people for badly executed manoeuvres, and by and large the next time they improve.”

Carrots make things worse

The instructor continued:

“I’ve often praised people warmly for beautifully executed manoeuvres, and the next time they almost always do worse.  Don’t tell me that reward works and punishment doesn’t.  My experience contradicts it.”

The rest of the audience agreed.  Worse still, when Kahneman looked at the data it was true.  Punishing poor performance works and rewarding good performance is clearly counter productive.  This left Kahneman a confused and worried man.

Hold that thought

We need to switch topics, I need to talk about dice for a second, but I will come back to it.

If you roll a couple of dice repeatedly the score you get changes over time.  It looks a little like this.

Variation in dice rolls

Why does the score change?  Because there are a whole host of random factors effecting it

  • The height you drop the dice from
  • How hard you throw them
  • The position they started in
  • The angle of the surface they hit
  • And and and

But the system (the two dice) didn’t change.  So although performance (the score) jumps up and down, nothing actually changed.

It is just noise.

Performance has far more to do with dice than carrots and sticks

Remember the story about the flight instructors?  Well a while later Kahneman learnt about variation and “regression to the mean”.  That may not sound like a fascinating topic.  But it held the solution to his problem.

Over the short-term how well a pilot lands his plane changes from one day, depending on:

  • How windy it is
  • What the temperature is
  • How well inflated the tyres are
  • How much fuel the plane is carrying
  • The time of day (how does the position of the sun effect visibility)
  • And and and …

If you look at the long-term average a pilot will get better as he develops his skill.  But in the short-term you might as well be rolling dice.

Random flight performance

Short term performance has nothing to do with carrots or sticks and lots to do with dice

But the kick in the pants is the “regression to the mean” bit

If you land your plane badly (or roll a double one) and your flight instructor screams at you, chances are next time your performance will get better.

If you land beautifully (or roll a double six) and he praises you, chances are next time your performance will be distinctly average.  It will “regress to the mean”.

Regression to the mean

Of course that has nothing to do with the behaviour of your manager / coach / flight instructor.  All he has managed to do is piss you off (strong but true).

Unfortunately the Israeli flight instructors didn’t realise that.

Did you?

If you enjoyed this post click here to receive automatic updates

Carrots, sticks, donkeys and regression to the mean

Read another opinion

Image by slopjop

Comments

  1. Hello James

    Another excellent post. If you have not already done so then I thoroughly recommend reading Kahneman’s masterpiece Thinking Fast and Slow.

    Sometimes I wonder how it is that a species with a cognitive structure like ours – simple, intuitive – has come to rule this planet. And then I remember that our expertise is in manipulating matter. in that domain we are Gods. Yet, the domain that really matters – influences the quality of our living, our lives – is the human domain. And in that domain we suck. Most likely because we have simplistic minds and those in power cannot help exercising power to feel good about themselves. After all if it is simply regression to the mean, which it often is, then what role is there for the traditional ‘carrot and stick’ manager? And where does he find his self-esteem?

    Maz

  2. Great post, James. I felt like I was reading about my 8-year old there for a second. :-)

    Thanks for putting a different perspective out there. Thought-provoking, for sure.

    Annette :-)

    • James Lawther says:

      As the proud owner of two children Annette, I can tell you that screaming at them is a hugely positive experience. It does absolutely nothing for them but makes me feel a whole lot better.

      Glad you liked the post

      James

      • LOL. Good point; it seems to have the same effect on me. I think our kids would get along well!

      • James,
        You’ve made me laugh! The same shows up for me and so once again I am present to the machinery of human being whirring away whether it lies in you or me.

        After the good feeling of shouting or telling them off. I find myself deflated and remorseful. Thankfully, that helps me to apologise and be a better father. Shout less, walk away more often.

        Maz

  3. James,
    What a great combination of a story and an analogy.

    Why do you think we have such trouble seeing that it is the system that we should fix and if we fixed it many of the ‘people’ issues we experience would just go away?

    Is it because we have built something, spent money on it, capitalised it?

    Adrian

  4. So the real quesion is “how do I change the mean line (for the better!). In the dice game, I do not think you can. In the human game, I am not sure if either the carrot or the stick can be used to move the mean line. Does praise alone actually provide long term improvement? Does yelling / demanding better performance ever move the mean in the right direction? Is real improvement really a function of clear standards + training + practice +++?

  5. “Psychologists will tell you that carrots are a far more productive strategy than sticks… Sometimes the psychologists are proven wrong. That is exactly what happened to Daniel Kahneman…This left Kahneman a confused and worried man.” “Well a while later Kahneman learnt about variation and “regression to the mean”.

    Ummm. I think you’ve taken too much license here. Kahneman tells the story differently. First of all he was not a confused and worried man about this. And it wasn’t a while later that he talks about the regression to the mean. He tells the flight instructor story to make exactly that point.

    Finally, you start your post with the provocative statement around reward being more effective than punishment when you said: “but the facts don’t always support it” But as Kahneman himself might say your inference “about the efficacy of reward and punishment was completely off the mark” i.e. not supported by the facts at all!

    • James Lawther says:

      Bill.

      You are right on both counts:

      1. In terms of the timing, I am guilty. As they say “why let the facts stand in the way of a good story”.

      2. The facts don’t support my provocative statement, not even remotely, that is the whole point of the post.

      The second issue though I’m afraid isn’t one of data, it is one of interpretation. I write in British English and you read in American English and we are divided by a common language.

      http://www.telegraph.co.uk/finance/2921790/Two-nations-divided-by-a-common-language.html#

      Sorry for the miscommunication, and thank you very much for your comment.

      James

Trackbacks

  1. […] love to add an extra fable to his collection–regression to the mean. Great article here on Daniel Kahneman’s work on the subject–detailed in his book, Thinking, Fast and […]

Speak Your Mind

*