On Thu, Feb 13, 2003 at 06:57:14AM -0500, Erik Reuter wrote: > On Wed, Feb 12, 2003 at 10:38:47PM -0600, Julia Thompson wrote: > > The bell curve is an example where the mean, the median and the mode are all > > the same. > > > > In a great many distributions, however, this is not the case. > > > > Take, for example, number of arms per person. > > > > The mean is going to be less than 2, because some people are lacking 1 or 2 > > arms, being left with 0 or 1. But the mode is going to be 2, because almost > > everyone has 2 arms. And if you lined up everyone according to the number > > of arms they have, and took the person in the middle, he'd have 2 arms, so > > the median is 2. > > On the other hand (groan), if you randomly sampled 100 people from the > population and determined the average number of arms from that group of > 100 and called the sampled mean x1, and then randomly sampled another > 100 people and found the sampled mean x2, and then kept it up for, say, > 10K samples, if you plot those data you would see a normal distribution > of the sampled means.
On second thought, for this case where less than two arms is so exceedingly rare, the sample size I mentioned is probably too small to get a decent Gaussian. In my example above, change 100 to 100K, and then it will approximate a Gaussian. -- "Erik Reuter" <[EMAIL PROTECTED]> http://www.erikreuter.net/ _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
