Last week numbers are all messed up too... the Wii sold 41,000 units, but WiiSports sold 55,000 units.

ioi is just doing some adjusting, it'll clear out. Also, the .1% extra is probably just from rounding.

Well, perhaps you shouldn't put so much faith in them, after all "sample sizes are small compared to professional tracking", but anyways I don't want to anger the beast, my original question was answered.

 Perhaps you should take a statistics class and then compare the hardware numbers against those of the proffesional traking services. VGchartz numbers tend to be very close to the other tracking services and compare well against the total shipped numbers given every quarter. Going on a website and insulting it is a good way to get a hostile reaction.

As people have already said, there was an adjustment this week in the past American data to better coincide with Nintendos shipment figures (he had actually adjusted down due to NPD totals only 1-2 weeks ago) Considering iois small sample had the Wii sales in America much closer to the truth than NPD I think he did pretty well. It's becoming very difficult to argue against iois hardware numbers for any major region. In fact I don't think there have been any discrepancies with data from the smaller PAL markets for a while now either. I personally am not so sure yet with PAL software [or some American software] but it is certainly getting more accurate.

ioi actually did not undertrack the Wii at all... NPD did... ioi adjusted to NDP last week only to have Nintendo back up his original numbers.

On the other hand, NPD's numbers are not gospel either since they don't track all of the largest retail outlets.  Most here, I would say, are more interested in the trends that ioi's data shows versus whether the rounding of a tenth of one percent should go to this console versus that console.  Or even to both.

One other comment, as mentioned before about statistics.  Small sizes can give fairly accurate results.  If you disagree with this, then you would have to ignore 99.99% (or higher) of all polls that sample only 1% (say, 50,000) out of a population base of 5 million.  Which for most polls would be a huge sampling of the population that the poll is about.