"In its calculations of the total software deployed during the year, open source software, freeware, and shareware were considered legitimate software and were not considered pirated. In calculating piracy losses, IDC counted this as paid-for software with a price of $0."

Lets do a scenario:

Average load is 5 packages per PC.

Bob has 5 packages, all open source free software.

John has 5 packages from Microsoft, all paid for.

Fred is unknown, but he has a PC.

Based on the known average, you could reasonably assume Fred has 2.5 commercial packages. But that's not what IDC method says.

Microsoft make $100 per package.

Hardware_units * Software_Load = Software_Installed

3 * 5 = 15

Software_Market_revenue / Average_Sale_Value = Units_paid_for

Software_Market_revenue = 100*5 + 0*5

Average_Sale_Value = (100*5+0*5)/10 = 50

UnitsPaidFor = 500/50 = 10

Ergo Pirated copies

= 15 - 10 = 5

Fred is unknown but yet assumed to be using ALL pirated commercial software. Even though he is likely to be using 2.5 packages.

The method is chosen to give a skewed figure. It's much worse than that, the survey is done on people who deal with commercial companies to give it a commercial software bias.