It's one thing to use cherry-picked data to build a case supporting one's bias...
But quite another to take data from another group and applying bad logic to support your bias. The stats, unless there is more in the NPD survey than reported, do not support the basis of the article. With the data presented, it states that in Q4'2011, people who never owned a smartphone were more likely to buy an Android-based phone than any other, while people who already owned a smartphone were... more likely to buy an Android-based smartphone, albeit at a lower percentage.
However, there are no listed statistics to back up the assertion that people who owned an Android-based phone were turning away from the OS to purchase an iPhone. Comparing two disparate groups and stating that it demonstrates any kind of movement away from the Android is low IQ at best, outright pandering to Apple at worst. The only conclusion you can take is what is presented, which is that Android is winning the mind and market share of those who are new to smartphones, while Apple, with a head-start in the smartphone biz, is holding 43% of all upgrades.
A more salient question that would actually answer the question of loyalty is to compare pre- and post-upgrade phone choices. Only if after looking at the data and see iOS with a larger share of the post-upgrade user base than the pre-upgrade, could you conclude that Android has a problem retaining users. Until that happens, all we can conclude with the data in this article is that the author fails basic comprehension of statistics. And if the information was present in the NPD survey and was not included in the article as the basis for the conclusion, it shows that the author fails basic journalism.