Last weekend I was at my nephew’s 8th birthday party. As luck would have it, my sister had placed out a few jars of candy with a pen and paper so the kids could guess the number of items in each jar. The usual suspects of jelly bellies, gum, and mints were all there, with a jar of pixie sticks thrown in for good measure. I sure didn’t see that one coming…but I digress.
While everyone else was focused on swimming, sangria and poolside chit chat. I was wondering just how smart a group of eight year olds really was. Fresh off the heels of reading the Wisdom of Crowds, I felt I was ready to put on my research cap.
The voting was done in plain view (a distinct break in the rules of what makes crowds smart), but since I didn’t run the experiment, I had little choice in the methods used. My sister assures me that next year the voting will be via ballot to avoid the bias caused by kids seeing other kids’ votes before they make their guess. You can see a sample of the Voting. I took each of the voting sheets and spent some time in Numbers doing a quick analysis of averages. Here is what I found…
In most cases the crowd did quite well. For example, the actual count of mints was 51 and the average of 13 kids yielded a mean of 53. The winner of this candy jar was Tom with a guess of 50. The actual number of pixie sticks was 111 and the mean of our group was 77 with the best guesser being Patrick who chose 100.
In these two the crowd did OK, but they were beat by individual ankle biters. However, note that the same person did not win both the pixie sticks and the mints. That is a key factor of the crowd. A single person may win a few, but over time, they will not beat the crowd’s average. There is no reason to think that a persons rightness will continue. The search for an expert is a fallacy.
Let’s talk gum. The actual number was 87 and our average was 94. Better than the best guess of 78 by Garrett. However, Garrett had his revenge in the final category of Jelly Bellies, winning handily with a vote of 704 and an actual count of 692. Our group failed miserably here with a mean of 336.
So what did I conclude? My main observation was how important independence really is. The only real area where the theory fell apart was in Jelly Bellies and in that area, my three year old, with my “help” voted first with 325. That vote clearly set the tone and a ceiling for what was “ok” and tarnished the results.
The other conclusion is my surprise in how well the group actually did in a few areas even with our complete lack of adherence to what makes crowds smart and small sample size. In fact, crowds seem resilient if nothing else. I plan follow-up studies with more tartar causing items and children. Perhaps throwing a few pets into the mix for some diversity.
You can see the full in-depth, highly scientific analysis in this pdf. (widsom of eight year olds)