@mrstickball:
First thing: I never said monthly reports aren't composed of number (4 or 5 obviously) week reports. So first half of your post has nothing to do with the debate.
But there is no evidence that a week will belong to given month if its Sunday happens in that month. And even more there is a contradicting argument as we know that it has no influence for monthly reports of two companies (ELSPA/ChartTrack and NPD). I'm not sure how exaclly it works in Italy.
If you are an analyst than you should know that there is a reason behind 4-4-5 cycle used by NPD/CH. Lots of comparisions are made with the same month from last year and thus having equal number of weeks in each January, February etc. is a good thing. We don't know excally if GFK uses that scheme but it would be logical if they did. Not necessary 4-4-5 (5-4-4 maybe just as good) but it would put this and last November at equal amont of weeks anyway (possibly with 5).
Don't play down percentages. It's not like we got real numbers from previous MS reports. Percentages are often all we can get and we have to be happy with them.
You are questioning last Novembers numbers but you said before that "At the end of 2008, we'll have LTD values for every region". So why didn't we get them last year?
The truth is that past numbers are more accurate as we get bunch of reports few months after a given period. I really don't remember any significant changes in old numbers but there were plenty for recent numbers. There still is a room for error in last year's November but its significantly smaller than for this year.
Basiclly you are saying that vgchartz numbers are worthless. And before you start attacking me I'm just saying that there probably recent X360 Europe numbers aren't too accuarate. And let's not play dumb here, recent numbers were often adjusted. But there is difference between my statements and saying that historical vgchartz numbers cannot be used as any reference point.
In the end I think that possibility that it's 5-week month vs. 4-week month comparision is very slim. Maybe GFK actually has 5-weeks Novembers but if so I do think that their 2007 November was also 5-week month. And even if it's not like this than I doubt that MS wouldn't point it out in its PR.
The other possibility is tham MS have calculated the numbers themself (ChartTrack and GFK can see a November differently) and in this case it's also vary unprobably that they would do it the way that makes them look worse (and didn't point that 5/4 week month out).