This is a short summary of an earlier more extensive analysis of this issue.
In short, no the iPhone doesn’t enjoy unexpected (magical) loyalty.
Essentially iPhone looks as if it has more loyalty than Android smartphones because we aren’t comparing apples with apples (pardon the pun). There are three reasons for iPhone’s higher loyalty than all android phones.
1) iPhone only plays in the premium end of the smartphone market, an area where users do use apps, and hence tend to be a bit more loyal because their purchases lock them into the platform.
2) People tend to trade up in smartphones, not down, so the premium end of the market has a bit higher loyalty than the lower end.
3) In this premium end of the market Apple is the market share leader, so in accordance with the Double Jeopardy law it gets higher loyalty.
So yes Apple has higher loyalty than all other smartphones, but its loyalty advantage over other premium smartphones (like Galaxy) is largely Double Jeopardy in action.
PS The “lock in” effect of iTunes and App Store is less than might be expected because buyers of premium smartphones are already so loyal.
Marketing Science 
Hi Bryon. We (BrandHook) have just completed a study that looks at the habit levels of the Apple ipad and the Samsung tablet – it looks like the ipad is more of a habit than the tablet. That is, the ipad is bought more on autopilot. We are watching their market share though because we suspect that like the phones, Apple will loose habit points as Samsung continues to disrupt. Love your work, Pipx
Dear Byron,
I read that iphone does have a higher loyalty, however the drivers of this loyalty are not what you would expect.
But is this finding based on reasoning, or have you been able to disentangle the drivers of the ‘loyalty’ in a causal model?
Best,
Roeland
Please see earlier more extensive blog post on iPhone loyalty.
PS By “causal model” and “drivers” you mean statistical modelling of correlations in a single set of data. Please see http://www.forecastingprinciples.com for why these predict so poorly, and hence aren’t used for much by scientists (other than for exploratory hypothesis generation). Also see the chapter in Nate Sliver’s excellent book on model over-fitting, social science’s dirty secret. Patterns that repeat (scientific laws) provide prediction, and hence explanation (see chapter one of How Brands Grow).
Thanks for your reply.
I read your book, it’s impressive! And I have a PhD, So I am quite familiar with the statistical challenges involved in disentangling these effects. That’s why I was curious for your approach.
Roeland
Here is another article you may enjoy/value. It’s a very short commentary from Ehrenberg on the general uselessness of best fit models of a single set of data, this issue was a major theme of his life.
Click to access 5461.pdf