Aielyn said:
Jabbamk1 said:
VGChartz has never been accurate anyway and always relied on external sources.
They use a prediction model to update the site and not a tracking model.
Anyone who actually takes this site seriously or looks at the numbers on the main page seriously are just fooling themselves.
The only official numbers come from Sony, Microsoft and Nintendo via actual market research companies like IDG, NPD, Gfk, Media Create etc...
Oh and people above saying VGChartz is being stopped by external forces or something you cannot be more than wrong as VGChartz don't even use their numbers. They just make them up.
|
People who want to claim that VGChartz is just "making up the numbers" need to stop making up the "facts" that they use to justify the claim.
VGChartz uses sampling data along with statistical analysis to estimate the sales. This is *exactly* what every other tracking company does. This is why, for instance, Media Create and Famitsu routinely differ by over 10% on many of their numbers. It's also exactly how polling companies work when determining public support for various things.
VGChartz don't have as large a sample as NPD, in terms of their American data, or as large as Media Create or Famitsu in Japan. But what they do have, which to my knowledge isn't true of any other videogame tracking company, is sampling across many countries in all of the major territories at once. This, in combination with the willingness to incorporate data from the other tracking companies to help refine their own data, is actually a strength.
Of course, statistical analyses will sometimes produce significantly incorrect numbers. That's the nature of statistics - you have a mean value, and then a confidence interval describing how accurate that mean value is. VGChartz, as with all other tracking companies, report the mean values, but not the confidence intervals. And even the confidence intervals are usually only 95% accurate - that is, 5% of the time, the real number is outside of that interval.
Anybody who thinks that it's possible to achieve even the level of accuracy VGChartz manages prior to release of other numbers through sheer "guessing" and "making up numbers" is kidding themselves - the probability that one could manage that level of accuracy (let's call it "within 30%, 99% of the time") by chance is diminishingly small. Are the numbers perfect? No. But then, neither are NPD numbers - they're estimates. NPD is more accurate for USA, but still not perfect.
In case you doubt what I'm saying, I'll note that I tutor university level statistics. I know what I'm talking about.
|
