I do not believe in benchmarking with motherboards, since the difference between the results are often within the margin of error, simply due to the motherboard not afftecting the perfomance that much in games and benchmarks. I believe the reviewer used lower resolutions since that wiuld make the fps difference larger compared to running it at 1080p with the highest settings.
Yes, you got it right. At 1080p you see that almost only the GPU has an influence on performance. To show the performance influence of the mobo you have to reduce the graphics load as much as possible.
Furthermore you're also right with your point that in some cases you're measuring standard deviation. Never the less it is interesting to see that for example in Super Pi one can see changes with every BIOS update.