You do realize that the site being used isnt going by sales but rather how many computers are actually being used?
I do know this: in the company I work for, PCs are turned over among staff every three years. Macs get turned over every seven. The PC users play with Microsoft Office, Access, Filemaker, and Crystal Reports. To bypass the $70/hr in-house chargeback to the IT staff, PC users will often call a $40/hr in-house Mac user (typically a graphic designer) to assist them with trouble-shooting files and instruction on how to use their own programs.
As I pointed out, web analytics can be used to show that the sites surveyed are visited predominately by Windows users, but to extrapolate that to stating that they can figure out the install base from those statistics is disingenuous.
And as I also pointed out, the author erroneously confuses the conclusions of the survey regarding install base with market share. The two are very different.