Here’s the scenario: You’ve launched your app, and you kept updating it, adding new features and supporting extra functionality. At some point, Apple releases a major iOS upgrade. Considerable effort is put into upgrading your codebase, and you reach a point where some features cannot be supported, unless users run the latest iOS version.

That’s the point where a decision needs to be made:
Should I keep supporting two versions of my app, one compatible with the old OS, offering limited support for new features, and a second one compatible with the new OS, or should I just abandon the old OS and support only users who’ve decided to upgrade?
As always, data comes to the rescue. The following table breaks down new installs and upgrades for David Smith’s My Recipe Book, and associates this info with the OS version of devices running it. For the purposes of this analysis it contains at the last 1000 user accounts created

There are a couple of interesting observations to be made:

In iOS, users generally upgrade to the latest version of the OS
Users who’ve decided to linger on to a previous version, are still buying new apps
There’s no difference in the buying habits of users who are left on a previous OS, they buy and upgrade apps with the same frequency as users who are running the latest OS version

So, dropping support for earlier OS versions can be justified, only on the grounds of a reduced user base, rather than the widely accepted myth that users who don’t upgrade, are also those who don’t generally download apps.

Source: http://david-smith.org/blog/2014/02/13/slow-updater-purchasing-habits/