The article doesn't say, and I didn't see a link to the report which might have provided useful information...

However, I can see two basic forms this survey question might have taken.

First: "which smartphone would you consider buying: iPhone, Android, Windows Phone, Blackberry, ..." (Or asking type-by-type "would you consider buying an X?"

Second: "which sort of smartphone would you consider buying?"

You can get wildly different answers by changing the form of the question. Providing a list and saying "which would you consider?" is probably going to get a lot of people saying "sure, I'd consider that", whereas asking without the prompt is going to produce a list of the devices with actual mindshare.