I was curious about statistics regarding devices 'in the field,' and all I could find were these figures from Google such as the ones you cite (55% on Gingerbread, 15% pre-Gingerbread), which I think should be tied to a very important measuring detail—they are percentages of Android OS versions hitting the Play Store during a two week period.
I think this is a deliberate choice of reporting statistic to make the Android device world look more modern. They could report on a number of other metrics[0], but from my observation, non-power users aren't frequenting app stores—perhaps they'll download apps a few times when the device is new, but after a while their phone is as customized as it's going to get. This issue is likely compounded after a few times of hearing about some app to try, and finding that it doesn't run on the dated OS.
Basically, I think there are loads of older Android installs in the wild, whose owners aren't going to the Play Store as often as people who just bought a new phone or tablet.
[0] I think something like access to Gmail via Android would be a better representation of the real market, and that figure is likely measured within Google. If not, it could be. I doubt the numbers would be very flattering... even the Play Store access metrics aren't so great to begin with.
I think this is a deliberate choice of reporting statistic to make the Android device world look more modern. They could report on a number of other metrics[0], but from my observation, non-power users aren't frequenting app stores—perhaps they'll download apps a few times when the device is new, but after a while their phone is as customized as it's going to get. This issue is likely compounded after a few times of hearing about some app to try, and finding that it doesn't run on the dated OS.
Basically, I think there are loads of older Android installs in the wild, whose owners aren't going to the Play Store as often as people who just bought a new phone or tablet.
[0] I think something like access to Gmail via Android would be a better representation of the real market, and that figure is likely measured within Google. If not, it could be. I doubt the numbers would be very flattering... even the Play Store access metrics aren't so great to begin with.