NPD to stop giving numbers to public?

[Sqrl]

NJ5 said:

Sqrl said:

Honestly I think its fair to say NPD was instructive in the early stages of the site for helping ioi refine his process but my take on what has been going on with the numbers as of late is that ioi has refined it down to a point where I don't think you can really say one is more accurate than the other. They are both estimates and within a fairly small margin and at that point its anyone's guess as to who is closer to 100% accurate.

But the good thing is that its also close enough for folks like us not to worry about it and quarterly reports will always be around and are considered hard numbers which are far more important than just another set of soft numbers.

edit: PS - woohoo I crossed 2k posts!

So NPD will start releasing quarterly numbers only? That makes it kind of pointless, that's when the hardware companies release their shipped numbers anyway...

This is great news for vgchartz, but of course not such good news for everyone else. Less information is less information, period.

PS: I wonder if the existence of vgchartz caused or affected this decision.

 

To be clear, the quarterlies I was talking about were the ones Nintendo/Sony/MS put out.

They are hard shipped numbers which means they are absolutely accurate numbers to check our accuracy against rather than NPD numbers which are just another estimate. I'm not saying that NPD data wasn't helpfull early on to get things refined but I think we are to a point where NPD data is innacurate enough that it can't help refine the process any more.

And for those who don't know, increasing sample size provides a diminishing return. So VGC doesn't need to get its sample sizes up to the level of NPD to be close to it in terms of accuracy.

I don't know how many retailers there are out their but if we assume 20k and say that ioi is tracking at 10% then the standard error is something like 0.64% compared to polls people accept for politics that are in the 3.5%-5% range.

If ioi is tracking down as low as 1% of that 20k things change and the standard error is about 7%.

Granted I am not exactly a professional on this stuff, and I am making some assumptions on the size of the population but those numbers aren't going to change a whole lot anyways since they are based on %s.