It sounds like you have some good experience there and I appreciate the insight. I try to use the same service station and fuel pumping protocol, but I don’t have any way to confirm the accuracy of the fuel dispenser. So far, the hand calc method is consistently 1mpg +/- less than the computer. Close enough for me really, and still not bad for a 4wd brick that weighs close to 5,000lbs and has nearly 500ft-lbs of torque. Never thought I’d be a fan of small-displacement, forced induction engines, but the last two new vehicles I purchased wore turbos and delivered impressive power along with fuel economy. Cool technology....
In terms of measuring the fuel, I think the problem is trying to refuel to the exact same amount after driving a certain distance, especially with the auto-shut off and different temps. Instead, we'd weigh a fuel cell before and after testing with calibrated load cell or scale. The procedure is outlined in different international test standards.
If I recall correctly, on my last engine project, the data from the ECU (spark ignited) was pretty dang accurate on our dynos and for in-vehicle testing. Of course, it depends on engine manufacturer and ECU manufacturer. Modern diesels are dead nuts in their metering, but those injectors are also very expensive.
In the end, no matter what method you use it's accurate enough to indicate performance, share with others, post on Fuelly, etc. Better to share the info, than search for perfection. I just have a pet peeve about definitive statements when it comes to test procedures. And engineers are not immune to it, as they are sometimes the worst offenders. It's like an engineering degree is a license to make unfounded claims.
