Tuesday, September 24, 2013

My "beer related" story on sampling.

Teaching "sampling" here in a market research course in Singapore. Why don't we run a census? First it costs too much. Second if you need to destroy what it is you're testing, youve got nothing left to sell.

Take a look at these Guiness cans. It would make sense to test these for bacterial infection before thsy get sent out of the plant. How? Open a can and run a bug test. So, how many cans to open so we can be confident in our result and sell the rest?
It was questions like these that led to the (very recent) discipline of inferential statistics.

Mmmmm. Guinness.

No comments:

Post a Comment