Oh - so he does know how to do percentages... why that one, and not the others? Ask if Microsoft can do percentages.
The 270,000,000 is the population... 5,700,000 is the sample. 16,000,000 is the number of discrete instances of malware found on the sample.
A legitimate question that should be asked is: "Can the results found in the sample be extrapolated to the general population?"
If it can, then we can extrapolate that 167,400,000 of those 270,000,000 computers were infected with at least one Trojan during the past 15 months.
So, is Microsoft's sample valid?
I little over 2% of the computers tested had some kind of malware.
That's not very many.
""Can the results found in the sample be extrapolated to the general population?"
Well, probably not. Those users who were tested are probably those who have legitimate software and keep it up to date. Those who do neither of those things might have a higher incidence of garbage on their machines.
But, what it really tells me is the whole malware issue is overblown if people are reasonably responsible with their computers and what they choose to install. I've always believed this to be true.