Was There Really a Hawthorne Effect in the Original Hawthorne Studies?


John List and I stumbled onto the original, never-analyzed data from the original illumination experiments done at the Hawthorne Plant. These studies gave rise to what is now known as the Hawthorne Effect.

We find that there actually wasn’t a Hawthorne Effect in the original data, at least not of the sort that you read about in virtually every introductory psychology textbook, where it is claimed that the workers’ output went up every time the lighting was changed, whether the change was to make the lights brighter or dimmer.

The Economist magazine has a nice piece on it.

Kenn Fong

The Hawthorne Effect must related to Heysenberg's Uncertainty Principle, which states that measuring a quantity affects the quantity you measure. It's like saying "I used to like going to Hawaii, but now there are too many tourists." The tourists go to Hawaii for the reasons people went there in the first place; it's a tropical paradise. But going to experience the tropical paradise changes it.



Sorry Kenn, that is an often stated but incorrect definition of the Uncertainty Principle. You are thinking of the Observer Effect, which, I agree, may be related to the Hawthorn Effect -- to the extent it actually exists.

Joe Hadfield

Love it love it love it.

Justin James

I'm not sure why you seem to think that you are onto something new here. The Wikipedia article yourself states: "Subsequent study of the results has found that productivity varied due to other factors such as the weekly cycle of work or the seasonal temperature and so the initial conclusions were overstated and the effect was weak or illusory." And points to prior work.


Garvit Sah

Interesting read. But intutively, aren't people expected to just push a bit harder when any experiment / observation is going on?


Actually, Justin James, it points to the same Economist article Levitt linked to.


maybe so, but if Hawthorne was watching, you would have changed your analysis


Don't you hate how data screws up a good story. . .


Garvit Sah: It's still worth it to test whether things really are as is expected.


Kenn: What you are talking about is called the Observer Effect, which says that the act of observing (or measuring) an object will slightly change that object. Putting a room temperature thermometer into a hot drink, for example, will cool off the drink.

Heisenberg's Uncertainty Principle is something else entirely. It is a _law of quantum mechanics_ that says, essentially, you can never measure everything about a wave at once. The more specifically you know where a wave is, the less you know about how fast it is moving.

The Uncertainty Principle is related to the observer effect, but they are definitely not the same thing.

Frederick Michael

You think that's strange -- death by horsekick in the Prussian army wasn't Poisson after all. That's right folks THE classic textbook, first ever, application of the Poisson distribution has big flaws. Just sum all the deaths for all the guardcorps to get the Prussian yearly totals and the sample variance-to-mean ratio is about 2.0 -- too high for a Poisson.

Maybe everything we know is wrong.

Doug Raymond

The Hawthorne Effect is a basic element of the canon for the training of middle managers, like the Hierarchy of Needs and the Japanese quality management system and so forth. What are we going to do without it?!?!?! Although it appeals to the sense of comic irony more than to the sense of moral outrage, the fact that we have abused each other with it for so many decades gives one cause to wonder what are we really doing in management training - perhaps that is the REAL Hawthorne Effect - we manage better for being trained, no matter what the actual training consists of. Would you believe that?

Hal Varian

See also

* Was There a Hawthorne Effect?
* Stephen R. G. Jones
* The American Journal of Sociology,
Vol. 98, No. 3 (Nov., 1992), pp. 451-468
(article consists of 18 pages)
* Published by: The University of Chicago Press
* Stable URL: http://www.jstor.org/stable/2781455

Daan Wynen

Is this data now available publicly? Or is it bound by some IP rights?