Representative Sample
Categories: Metrics
One pan of gold. Is that enough to let you know whether or not you want to lease the 50 acres in the Yukon for $100,000 a year minimum plus 15% of the gold you mine? Maybe. If your one pan comes back with 50 colors of gold in it, then sure! If not, then maybe you need a bigger n-sample.
The key here: if you're going into unexplored territory, whether it's testing to see if a market is ready for 3D glasses or Smell-O-Vision or driverless cars, you have to test a big enough sample so that the results that come back are statistically meaningful. Like, if you mixed up a sample massively and intelligently, you could sample only a very small portioin (like 1%, maybe less) and get a good feeling for the market...whether it was ready or not.
