Statistical Ties? Do they really matter?

[Barozi]

izaaz101 said:

Barozi said:

izaaz101 said:
Within 1 or 2% I consider it a tie. For me a "win" has to break the 3% barrier.

hmmmmm well I could be wrong here, but AFAIK if the margin of error is +/-10% then every percentage within the range has the same probability.

Which means if NPD numbers and VGC numbers differ by 10% both have the same probability to be right. But that also means that even if they report the same numbers the probability doesn't increase.

I'm not disagreeing - it's just a mental thing for me.

alright :)