I definitely see the value of testing, I just don't have a realistic way to accomplish it. I live in the heart of a city of 7+million people so finding a good test route to do ABA testing or even an empty hill to do terminal velocity test is pretty much impossible. Combine that with the fact I don't have a garage and do all my modifications streetside and it's even more impractical.
My personal approach has been to scour the site for mods others have tested on my car and replicate those mods (with whatever tweaks make it easier for me). MetroMPG is famous for testing his mods and most of mine have been direct copies. I try to do good how-to write-ups to be helpful to others who want to try these mods and add innovative ideas for modification to the community.
In the case of the belly pan MetroMPG did do some ABA testing but it was too annoying to put it on and take it off and the gains seemed too insignificant from the limited test so he didn't continue with it. This is the only mod I've done where there wasn't solid groundwork already laid for my specific vehicle.
I think it would be cool to know the % improvements of each mod but I just don't have that data. I know it's not scientific and people aren't fans of tank to tank data but my personal approach is that is the real world test measured by my wallet. In a way I'd rather see the MPG results tank to tank than on a spreadsheet because I actually pay for the tanks.
I'm just doing what I can - ABA testing is not in that realm but I hope my posts are found helpful to others anyway. And in the end it's a fun hobby to mod, hipermile, and post so I don't have any regrets