In any case, software numbers are often proven to be off when the npd report is released and gaf starts getting the numbers.

You *do* realise that NPD uses the exact same methodology as VGChartz, right? They collect a sample, and report on numbers by scaling it up.

And the great irony in all of this is that NPD doesn't even do people the service of telling them how many retailers they use in their monthly sample, so despite the protestations of so many anti-VGC people, there's no way to know whether VGChartz or NPD has a higher sample size.

Simply put, if you assert that the NPD numbers are "right", that where NPD disagrees with VGChartz it must be VGChartz that is wrong, then you're demonstrating your complete ignorance of the situation.

I have no doubt that NPD has way more resources and is more accurate than vgchartz. Which one would you trust more?

Let's suppose that VGChartz had 2000 retailers in the US, and NPD had 5 times as many, at 10000, just for the sake of argument. Then rather than a 95% confidence margin of error of 2.2%, it's now just under 1%. In other words, while "more accurate", it's not exceptionally accurate.

And you have "no doubt" that NPD has "way more resources"... but you have no evidence, and no reasoning to justify the claim... at least, none that you have provided to us.

So here's your challenge - provide solid evidence that NPD has more resources than VGChartz with regards to video game sales data collection.

If first parties didn't think that NPD was the most accurate source of data for NA, they wouldn't be quoting the numbers everytime NPD releases their report. I wish someone had a collection of the comparisons of NPD vs VGChartz numbers. VGC's purpose is to fill the gap for system wars for when official numbers are not available. I still recall the sensationalist "DS becomes best selling system of all time".

So I call for solid evidence, and you give me "publishers quote NPD"? Really? How does that prove *greater* accuracy. All it proves is that they find NPD to be accurate enough for their purposes. It says nothing about NPD vs VGChartz, it says nothing about resources, and it says nothing about the actual level of accuracy.

But more than that, you refer to "official numbers" once again - which proves that you haven't been paying attention to what anybody else has been saying. There is no "official number" available. NPD is an ESTIMATE. Same with Chart-Track, Media-Create, Famitsu, etc. They all use estimates, and it's the publishers' PR arms that use those numbers for their press releases.

I'm going to challenge you, this time, to do something much simpler. Just repeat after me: "NPD does not provide official numbers".

EDIT: By the way, here's Pachter revealing something about his private talks with publishers regarding NPD numbers: http://www.ps3center.net//news/5235/pachter-npd-can-and-has-been-wrong/

"... However, analyst Michael Pachter has said that whilst console manufacturers usually tell him that the NPD numbers only have a 1-2% margin of error, that is not the case for some software sales. He says that publishers have told him that the "NPD does a crappy job.""

The best part is that this quite neatly backs up the suggestion that NPD probably has somewhere around 2400-9600 retailers in their sample set, as 1-2% error with 95% confidence is the expected amount at that sample size. Assuming ioi isn't lying about the number of retailers he gets data from (and I don't see any reason to disbelieve him), and that US retailers make up somewhere around 40% of the set, this puts VGChartz in very much the same ballpark, statistics-wise, with NPD.

I don't generally trust Pachter when it comes to anything that is even remotely resembling PR, such as comments about how certain platforms or games will do. But I see no way that the comments quoted in the link above could operate in such a manner, and thus see no reason to distrust him here.

NPD releases monthly numbers 10 days or more after month's end: this alone gives them more time to refine data, a little more for last week, a lot more for the previous ones. Add to this that doing a monthly total naturally filters out variance better than weekly numbers. BTW, giving monthly numbers NPD is able to adjust at least 3 weekly numbers a month before even releasing them added together in the monthly total. If you compare monthly VGC numbers at the same time NPD releases them, so with VGC having the same time as NPD to adjust the data of the first three weeks of the month, they should be percentually closer, on average.

I keep hearing "blah blah blah, don't matter numbers, method or validity, I believe different then I'm right". In other news, "my sacred escriture says my god is the only one, so he is''.