Quantcast
View Post
ioi said:
We are not adjusting any numbers. We feel our numbers are good, based on sound methods and from a large and varied enough data source to be statistically sound.

Let me make a simple point.

Say there are two retailers - retailer A and B.

One tracking service gets data for 5% of each. Retailer A says the hardware sold 5,000 units, retailer B 2,500 units (from this 5% of total stores). This tracking service estimates that a total of 100,000 units were sold at retailer A as a whole and 50,000 at retailer B by simply scaling up. So they say 150,000.

The second tracking service gets all data from retailer A and none from retailer B. Retailer A actually sold 105,000 units in total (not the 100,000 estimated by the first tracking source). Past data shows that retailer B as a whole is 75% the size of retailer A, so they estimate 78,750 units for retailer B. Giving a total of 183,750.

In reality the number is 105,000 for retailer A and 55,000 for retailer B - 160,000.

Which method of tracking gave the best results?

Of course I have purposely skewed the data to support my point, but it is not inconceivable that using many small sources of data could be better than fewer sources of complete data.

Ah! excellent, so i got it right. (see comment about pies near the top of the page)

 

@origin, ioi isn't saying he is keeping VGC numbers because of a principal or because he wants to be seperate from NPD, he is doing it because for the past few months his data has shown to be better for the 360 and DS than when we adjusted to be like NPD.... the fiscal quartely statements (Nintendos was combined to a half year statement) showed that there was a stupidly massive discrepancy between the number of DS shipped and what NPD had for sales, there would have to be dozens of DS sitting on every store shelf to acount for it, the 360 discrepancy was even bigger, though 360s have a much better retail supply than DS it was clearly not as much as the difference between shipments and NPD sales.

Maybe that is why NPD were stopping releasing public data this month, because MS and Nintendos own sales numbers (which they seemingly don't release) showed NPD to be off.