A flawed statistic - but amazing Apple can't even claim the one stat that's biased towards them
A shop sells 100 chocolate cakes a day, and 100 chocolate cakes with cherry on top. It also sells 101 fruit cakes. Media claim, fruit cakes most popular!
In a parallel universe, the shop has decided to relabel the fruit cakes by those with 16 raisins, 32 raisins and 64 raisins. The sales remain the same, but it reports now the three varieties of fruit cakes as 50, 30 and 20. Media astounded that now, chocolate cakes are more popular!
"Best selling single device" is a very poor statistic. I would dispute we can tell the most popular just by looking sales due to the problem shown above, or even if the concept is well defined at all. Furthermore, it all changes depending how individual models are labelled. The only relevant stats are by platform (where Android massively leads), or company if you care about their success (where Samsung massively lead over Apple, and Nokia in fact are 2nd). For most people (consumers and developers), I'd argue platform size is all that matters.
This is the same kind of problem as FPTP in voting systems - but worse, as the arguments in defence of FPTP don't apply here. People can argue that there is no perfect way to vote (due to the voting paradox) so we might as well stick with FPTP, but here, we don't have to pick "most popular individual device" at all, as there are better things to look at (platform sales). Also they can argue that the circumstances where FPTP fails badly are often hypothetical, but this is a very real world example of the problem: there are thousands of Android devices, and most companies have loads of models, whilst Apple only have one model to choose per generation. So Apple phone buyers will all be buying that one phone.
"And don't forget - especially if you're a fan of neither Apple nor Samsung - these handsets only 24 per cent of world smartphone shipments."
Indeed, which is further evidence why it's a poor statistic. Let's take things to extreme - imagine 99% of people buy Samsung Android phones, but these are all spread evenly across a large choice (more than 99) of similar phones that Samsung offer. It really takes one hell of an RDF to claim Apple as most popular, because their 1% share is from a single model. It also means that Samsung are penalised for offering more choice to consumers!
But this story is still interesting. The media only cling to this way of measuring, as it makes Apple look best. It's telling that Apple lose this stat now, even though it's massively biased towards them, with them only having one phone per generation. They are now so unpopular, that even one single device out of thousands outsells all the latest generation of iphones. What will the media do now? Will they finally give up on the Apple obsession? Or switch to "Oh, but the next iphone will sell more, honest"?
It's possible that the Nexus 4 will cause Android sales to rise further (good for Android), and also cause sales to spread more evenly between Samsung and LG. This is also good - it's more healthy competition in the Android market. But the effect could be to make Apple look "better" when the media quote the stats of best selling individual device, because the S3/S4 sales are now shared with the Nexus 4. So despite even further domination of Android, the media will be spinning this as a win for Apple!