When I see reports like this one I have to step back, take a deep breath and say, "oh well" when it comes to speed tests, especially, those done for mobile phones.
You see, not only are all speed tests not created equal. But for mobile, it's a whole new ballgame as a series of factors impact the throughput and speeds one will receive, making nothing a really fair test, but really a snapshot in time for that specific location at that moment.
1. Distance to the cell tower
2. Current load on the network
3. Speedtest site used
4. Load on the route to the speed test site itself
5. Height/floor of a building you are testing one
6. Possibility of in building wireless DAS system
7. Potential of a network operator installing an in building pico cell.
All of these factors can change the result one receives when it comes to speed, and thus, mobile speedtests are never fair fights. As a matter of fact, they're in the end, nothing more than a diaper derby.