Well, viewing this objectively and pedantically as practical in a few minutes, the box works out to 0.39m2 against an estimated (by one source) 2.5m2 for the car, a 16% increase in the aerodynamic component of losses assuming the same CD as the car, but I'd bet it's somewhat less. According to one detailed but unvalidated mathematical

model of Kona losses, aerodynamic drag represents 46% and 56% of the total losses at 90 and 120 km/h respectively. So, the additional drag might represent an additional (.16 x .46 =) 7% and (.16 x .56 =) 9% more total losses.

The measured differences are 4% and 7% more kWh respectively, where the estimate above says 7% and 9%.

But each measured

*pair* of numbers have a relative potential truncation or rounding error of around 4% and 6%, assuming that a multitude of other practical influences have been avoided while driving. That leaves our measurements indicating a very loose 0-8% and 1-13% against an estimated 7% and 9%, the latter within the margin of that error.

So, it's not unreasonable to suggest that the measurements are

*not* wildly off base, especially if the actual box Cd is better than the car.