| Kasz216 said: The question totally misses the point in general. It'd be like asking whish number is higher .99999~ or 1 .9999999~ = 1 despite the fact that 1 looks higher. They're the same number... period. Numbers given with a margin of error aren't "real" numbers. They're the middle point expression of a possible range of numbers. To say statistical ties don't matter is basically to say what NPD does is meaningless since you are disregarding the foundation on which said numbers are built. It would be like saying you don't think the basic laws of physics are important but then taking physisits findings serious. It's incongruent with logic. |
Uh, no. "Statistical tie" is the more generally used equivalent of a "statisitically insignificant difference" (the statistics term). The numbers are not necessarily equal. Rather, their ranges overlap. In statistics, the probability that the number you got is actually the exact correct value is infinitesimally small. Ranges are much more valuable, as they can represent a set of values that there is a 90%, 95%, 97%, etc. chance that the real value is in. Now, if we assume there is a + or - 5% margin of error with 95% confidence, a difference of 3.84% is very small. It's a fact that there is a statistically insignificant difference, and while it's true that it's more likely that the PS3 won, the chance that the X360 won isn't that small. It'd take some integration to figure out the exact value without the data, but it's probably somewhere between 20-35%.
