It does mean you can't meaningfully compare the two.
Are you saying that if people started installing Android onto iPhones (if it were possible) all of a sudden the hardware failure rates for iPhones would sky rocket?
All this says is that devices running Android fail on average more often than iPhones – for whatever reason (regression to the mean, one designer and manufacturer vs many, lower priced devices, and so on). That’s a perfectly valid comparison.
The comparison is invalid because iPhone/Blackberry devices are premium high cost devices with high quality control.
A direct comparison will only work against other premium high cost devices with high quality control.
By comparing every Android phone on the market (from cheap to premium) to only premium devices you are always going to get a skewed result.
All this says is that devices running Android fail on average more often than iPhones – for whatever reason (regression to the mean, one designer and manufacturer vs many, lower priced devices, and so on). That’s a perfectly valid comparison.