large snip

NO! To get better than 95% probability that it's not random chance,
the required percentage correct drops as the number of trials
increases, so that while you need 9 out of 10, you only need 15 out
of
20, and this reduces to not much more than 500 out of 1,000 (for the
real obsessives!).

Yes, it's been pointed out that it should be 15 out of 20, not 16,
for
better than 95% confidence. That's fine, rules changed accordingly.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

I cant believe you just said that : 500 out of a 1000 is proof
positive with a 95% confidence level.

to any normal person with even basic O level maths, its 50%, random
chance.
We all did the silly dice throwing and looked at the results when we
were at school.
Later re-examined in college stat's class, it was still 50:50.

Unless your current job is Advisor in Mathematics to the New Labour
Education department. or the Treasury -- Its the only rational
explanation.

I suppose anyone doing your test that scored 6 out of 7; is still
random chance with a 35% uncertainty.

I cant believe you just said that : 500 out of a 1000 is proof
positive with a 95% confidence level.

He didn't. He said not much more than 500 which is more than 500 just not much.
He is quite right. Small percentages in difference mean a great deal with large
enough samples.

On 15 Jan 2004 00:01:13 GMT, "chris"
wrote:

large snip

NO! To get better than 95% probability that it's not random chance,
the required percentage correct drops as the number of trials
increases, so that while you need 9 out of 10, you only need 15 out
of
20, and this reduces to not much more than 500 out of 1,000 (for the
real obsessives!).

Yes, it's been pointed out that it should be 15 out of 20, not 16,
for
better than 95% confidence. That's fine, rules changed accordingly.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

I cant believe you just said that : 500 out of a 1000 is proof
positive with a 95% confidence level.

Your disbelief is well founded - I did not say that. Read it again.
--

Stewart Pinkerton | Music is Art - Audio is Engineering