“It isn’t statistical evidence because the sample size is too small”
I didn’t posit it as statistical evidence. I observed the numbers, and having known kids who have attended the bee, I know a little (albeit indirectly) about what it takes to win such an event.
The poster I replied to gave these numbers on school makeup:
School Type:
192 public (67.13%)
38 private (13.29%)
36 home (12.59%)
14 parochial (4.9%)
5 charter (1.75%)
1 virtual (.34%)
I’d say that this is not a statistically invalid population...but that’s as far as I would go in describing it.
It is invalid because it is not a random survey. It is only a survey of students who are very smart. If you wanted a random survey you would pick home-schoolers and publicly schooled children from around the country at random and then test them. With a distribution size in the millions you can't draw a conclusion just looking at the end of the tail to determine if home-schooling is more effective than public schooling at teaching spelling. This is like saying that men are many times better at governing just looking at the composition of the House and the Senate--which again is hardly a random sample.