The test was done using an iPhone 4 and a Google Nexus S. All websites loaded were Fortune 1000 company websites, and the sites were loaded more than once, and on different days. The testing was done over WiFi, so carrier discrepancies can be ignored. There were multiple devices running the tests, and the median time of each site across the devices was used as the result.
Blaze does acknowledge that a perfect page load test is practically infeasible. The amount of unique variables that factor into browsing performance is staggering, and trying to standardize each one would be a waste of time and resources. So, while this isn’t definitive by any means, it definitely favors Android, and there’s something to be said for that.