No dice.
Even re-timing the recordings of the runs to include the A run I had to omit earlier, the wheel skirts do not come out ahead:
In this case, adding that previously thrown out non-skirt "A" run causes the standard deviation for the "A" set to be higher than the difference of the average of the A & B sets. Which means I can't really claim any statistical difference between the two sets.
In conclusion! Here are some possibilities:
1) The skirts, as I constructed them, really don't help even though they appear to
show improved airflow in the tuft testing video. This could be explained because they increase frontal area in order to reduce Cd.
2) The testing was flawed. The amount of variability, particularly in the A (non-skirt) set, is pretty high.
One thing I can say for sure: trying to do accurate testing is hard and time consuming!
I think it's also fair to say: the front skirts (as I made them) certainly don't appear to be a slam-dunk aero band-aid
for this car. Based on these results, it's probably reasonable to say they're so close to making no difference one way or the other that time would be better spent on other aero mods.