That
does sound like a lot of starter failures! I don't know that I start my stuff 10 to 15 times a day, but it's probably at least half that (unless I'm engine-off at red lights & such, then it's probably that much), and the starter in the Gnat has been replaced ONCE since I've owned it - and that's been several years now. Can't remember EVER replacing the starter in the Jimmy, and I've owned it even longer (and driven it much more)...
Oh, mileage for A-B-A testing. Do you mean how many miles long should the test course be? I have one that's about 4.5 miles long & it works pretty well. You could probably get by with less, but I would think you'd want at least a couple miles each way. Do you have cruise control? Because like Darin (MetroMPG) sez, that's a good way to remove the variable of your right foot from the equation.
Some things to look for on a test course;
> Fairly level and flat
> Good landmarks for where you start/stop your test going each way
> Places to turn around far enough beyond your start/stop points for acceleration & settling-in to your target speed
> A speed limit & traffic volume that will not interfere with your tests (much)
> Conveniently located (you don't want to have to drive over half an hour just to get there)
As for the start/stop landmarks, I've found out the hard way that if possible, you want them to be in areas where the road is pretty straight & level. Because if it's not (like if it's climbing, as in one of my early ones, you can see the MPG dropping as you head up that hill.) When that happens, the time you read your final "current" mpg average is critical. If the number isn't changing at the moment you read it, it's gonna be a lot easier to have confidence you got a good reading.
(I'm a retired land surveyor, so I'm all about "good readings")
