Surely in this case lab tests are the only reliable metric to use? I know that when comparing manufacturer's wider economy claims, it might be valid to cite real world experience. Because those general claims can be affected by a raft of variables which can't necessarily be accounted for in a fixed laboratory environment.
But if you only have one variable to test - if the ethanol content affects efficiency - then surely a lab test is the best and most fair way to judge that?
Place the engine on a dyno. Test it's output on E5 fuel. Then test again, with the only difference being changing the fuel to E10. Measure the difference and you know precisely how the ethanol content affects the engine output, and therefore, it's efficiency.
Surely "real world" testing is untrustworthy, because as has been pointed out, the switch to E10 has coincided with the autumn/winter seasons. Not only that, but rumours of reduced efficiency have bene circulating since before the switch. Measuring a car's economy is a tricky thing to do accurately, and can be affected by subjectivity - if lots of people already believe E10 is going to lead to reduced economy because that's what they've heard, it doesn't take much for them to start blaming that if they do see their economy drop even a little.