Alright! Alright! Calm down!
Before the iPhone fanboiz and their opponents start getting too aerated, have a look at the link to the original press release given on the Register article. Then think about it. Three problems are immediate:
1) What's the baseline? In other words, from what figures are these percentages made? NPD's last survey? They imply that, but don't say. Without some sort of baseline figure, it's like saying "Bill runs faster than Fred by 15%". Well, how fast does Fred run? He might be a crippled WWI war veteran, for all we know.
2) What's the sampling, and how is it chosen? NPD says: "sales data based on more than 150,000 completed online consumer research surveys each month". How are these online consumers chosen? Is it a fair sampling, or is it skewered towards customers of Blackberry outlets and against those from iPhone outlets? NPD leaves us clueless. Is 150,000 a fair sampling of wireless mobile phone sales in the US, or is it the equivalent of asking four dog owners whether their dogs prefer Winalot and extrapolating from that? We can't know.
3) What's the margin for error? All serious survey companies - NOP, yougov etc, will include a margin for error (in their case + or - 3%, but it can be higher or lower). I don't see it in the NPD survey.
You don't need to have a mathematical degree in statistics to spot these problems - I certainly don't. You just need to have read The Tiger That Isn't by Michael Blastland & Andrew Dilnot (quite cheap on Amazon) to start making real sense of surveys, figures etc regurgitated in the media, including the Register, to see through them clearly.
I have no particular axe to grind one way or the other on the wireless phone issue. NPD's survey and analysis may well be correct. It's just that we have no way of knowing.