A lot of people seem to be misreading the GfK sales comparisons.

The original story looked at the publicly released sales totals through for UK in January and the same figures for July and did simple subtraction to estimate sales for the six-month period. However, it failed to take into account the possibility of any intermediate adjustments made in the numbers.

It is good to hear that some shipment data was also used in making these adjustments. Because trying to determine sales is not an easy task. Just as GfK about its presidential polling. It showed the US Presidential race at 1% three weeks before the election while every other poll but one showed at least a 3% margin and most showed larger. (Only a single tracking poll showed the margin that close for one day. For more on this, visit RealClearPolitics.com)

The reason I bring this up is not to discuss politics or belittle tracking firms. Rather, it is to show that these numbers generally have a "margin of error." The polls we take for granted usually have several percentage points -- on either side -- leeway for a 95% confidence interval. Using that, 3% (a typical polling margin of error) on 30M is 900K. In other words, the numbers for all services tend to be "approximately right" and "exactly wrong."

Mike from Morgantown

PS -- I will refrain from my normal discussion of potential sample bias because I don't know anything about GfK's sampe composition (unlike NPD's which does not include Toys R Us and Wal-Mart which are places much more likely to sell non-HD systems).

What 90%? Do you know who they survey exactly?

Ahahaha, Kowenicki, how you slay me.

"Almost becoming niche".

Such humour.

does kowenicki know the meaning of niche?

Stupid comparison; the concept of the "home console" itself in Japan is appearing niche these days, whilst the handhelds reign.

But it only comes down to a few thousand sales gone either way.

Last summer, IIRC both PS3 and 360 were (whilst closer in pricing) doing ~ 11k weekly.

This summer, (with the pricing so in favour of the 360) 4k of PS3 purchasers have swung the other way, taking it to 7k vs. 15k weekly.

So suddenly PS3 and 360 have gone from home console weekly marketshare parity to a share of 2:1 in favour of 360.

With a PS3 pricecut, several thousand could swing back the other way, regaining parity, if not going on to turn things in PS3's favour.

In other words, it's not a crisis and it's very rectifiable for Sony.

Not in the UK it won't, PS3 is fighting a losing battle here, 360's are everywhere!

Agreed. In most stores you'd be lucky to find 2 blocks of PS3 shelf space against 8 blocks for 360 and usually around 10 for Wii.

On top of that Forza 3 is gonna have loads of advertising and that will perk the interst of alot of Europeans.

With 90 percent sampled, it is much less likely that some big retailer which could sway true results is being omitted. (This is much different than NPD not having Wal-Mart and Toys R Us data in its samples).

However, since it is sample, it is always possible to get a "bad sample" -- though the likelihood of that happening repeatedly is highly unlikely.  But such a situation could lead to strange numbers for a certain time period because of (a) the bad sample or (b) the need to correct for a previous bad sample.

This all being said, there have always been some strange numbers coming out of GfK in the UK -- meaning that there might be some data issues or some gamer issuses.

That being said, I would trust the cummulative numbers much more than the "derrived" six month numbers,

Mike from Morgantown

First time in a while i've seen these adjustments favor the 360 over Wii and PS3. Normally PS3 was the one creeping up in marketshare.