think it that way:

on a clear day you can read font X 10 words in under a second. font Y can be read 12 words in under a second. both are awesome, and a change would not improve anything.

on a night rainy day with little illumination, font X can be read at 7 words per second and font Y can be read at most on 3 words per second. Making font X clearly superior to font Y.

yes they would, if a font is more readable, it's more readable. And you can test it measuring recognition time and distance. And you can test in the sub optimal conditions mentioned, just like you could test readability on sub optimal substrates, resolution, degradation etc. It's not easy, but it's testable. i would like to see these articles accompanied by some in car photos of the signs being visible, where their conclusion is shown.