Experimental design: generalization questions

Why does experimenting on small groups not generalize?

``Eight out of ten people at the supermarket checkout with a dozen tins of MoggyScoff in the shopping basket said their cats preferred it''

Why does this experiment not generalize (two reasons)? Is it always the case that generalization requires large experimental groups? Does having a large group always improve generalization?

©1994-2003 Kevin Boone, all rights reserved