Sorry about making a new thread, but the computer I'm on doesn't let me reply in existing threads (but I can create new ones O_o), otherwise I would have put this in the official NDP thread.
Firstly tolerance / margin for error: NPD is an estimate, VGC is an estimate, therefore there must be a built in tolerance for differences in numbers reported between NPD and VGC. From my perspective if the VGC and NPD numbers are within 10% that's a tight correlation, and no one really has justification for criticism. The general view is that NPD is more likely to be an accurate estimate than VGC, so a difference in numbers outside tolerance would suggest VGC numbers may need some adjustment.
With that said, lets look at June 2010 NPD and VGC data as reported here http://gamrconnect.vgchartz.com/thread.php?id=113069&page=3 thanks WilliamWatts for neatly putting NPD and VGC into a single post.
So here's my assesment of accuracy (my grades are purely arbitrary based on my opinion on the significance of the difference, any number within tolerance automatically gets an A):
% diff Grade
PS3 -29.0 D
X360 0.3 A+
Wii -9.7 A-
DS -5.6 A
PSP -31.2 E
RED DEAD REDEMPTION 360 -19.5 C-
SUPER MARIO GALAXY 2 20.2 C-
RED DEAD REDEMPTION PS3 -21.5 C-
NEW SUPER MARIO BROS. 36.4 E-
JUST DANCE -10.5 B+
HW overall grade B+
SW overall Grade C-
As I see it Software is in for some significant adjustments, and not just the titles with numbers published by all accounts. On software it seems like week by week and even month by month sales should not be relied upon, but lifetime sales (with the regular ajustments that happen during a title's life) should be a reasonable representation of likely sales (accuracy would be greatly improved when publishers release game shipment data for example).
Hardware, the only adjustments I see as being needed are PS3 and PSP to be adjusted to come closer to NPD. That's not to say I think VGC should simply adopt the NPD numbers, given we are talking tolerances and estimates. Ideally VGC and NPD should be within 10%. But when they are not within tolerance what should be the adjustment? Should VGC simply adjust numbers so that the difference is 10%? My response would be no. After all NPD is also an estimate so its numbers also have a margin for error, only it will be a smaler MfE than VGC's. The balance of probilities is that the true number of sales, will fall somewhere between the VGC number and the NPD number.
So factoring the error of both sets of data, and depending on how heavily NPD data is weighted for the purposes of determining adjustments, I think VGC's PSP numbers for June could be adjusted up to betwen 107 - 113K and PS3 number adjusted up to between 268 - 283K.
“The fundamental cause of the trouble is that in the modern world the stupid are cocksure while the intelligent are full of doubt.” - Bertrand Russell
"When the power of love overcomes the love of power, the world will know peace."
Jimi Hendrix