I wonder if any of you can refresh my memory on who made a study of this subject:
I remember reading once that a mathematician went to a fairground and saw a stall where you had to guess the weight of a cow and, if you guessed correctly, you could keep the cow.
What he did was to total up the guesses of all the other entrants and then found the average weight
and he wrote that figure on a slip of paper. When the cow was weighed his answer was something like 1LB off the actual weight. He studied this again and again and came to the conclusion that a team of people could always form a better guess than any individual, if they used this technique. So long as the guesses were not outrageously off the target (for example it would be ridiculous to say there were only three chocolates in a jar if there were clearly more than 20).
I just wondered if anyone knew of this and who the person was that did these little tests?
Last edited by cray (2006-10-10 05:26:15)
That's an interesting theory. I don't think it would work with all situations though.
If you asked a group of people to estimate something where the answer was counter-intuitive, then they'd all go too high or all go too low, so having more people wouldn't help.
Why did the vector cross the road?
It wanted to be normal.
By taking the mean and the standard deviation a couple of guessers could get together and do better than a single one. How much better? That would depend on the sd. But would that increase the expectation of the guessers? Probably not.
If it ain't broke, fix it until it is.
A number by itself is useful, but it is far more useful to know how accurate or certain that number is.