Yesterday was 20.10.2010 on the European calendar, and it was declared by the United Nations to be World Statistics Day. In honor of this, the Statistics Department at Miami hosted an event that included two talks and a reception.
We did our best to publicize the event, but we really had little idea about how many people would attend.
So, about 45 minutes before it started, I sent the following note to our faculty:
In the spirit of this international day of statistics, please send me your confidence limits for the number of people attending our stat day event (I will count at roughly 4:30). For instance, if you think that no less than 100 and no more than 130 will come, your interval would be (100,130).
Whoever has the narrowest interval that captures the true number attending is the winner. No word on a prize, so honor will have to do.
In total, seven intervals were put forth:
(20,50)
(35,55)
(50,70)
(50,75)
(50,150)
(55,65)
(57,70)
The actual number of people attending: 41.
The second interval in the list above was the winner, because it was the narrowest that covered the true value. Overall, only 2/7 covered the actual value of 41 which suggests that as a group we were overconfident in our ability to accurately predict. Our intervals should have been wider to accommodate our uncertainty.
At any rate, I enjoyed the event quite a lot. The talks were excellent, the food good, and the fraternization heartwarming.
P.S. To be technically correct, I should have referred to “prediction limits” instead of “confidence limits” in my e-mail.

