How did you measure the elapsed time?
I have been collecting coast down data for about two months now, and I'm starting to reach the conclusion, that the GPS is not accurate enough to really measure the deceleration properly.
I also noticed that the coast down data at the Instructables looks a little fishy after really getting into the physics and math of coast down tests.
If you take the time to really process the data, and calculate your "change in velocity" or dV/dT (delta-V over delta-T), then you will notice that the rate of deceleration at high air drag speeds, will give you constant change in deceleration over time. In otherwords, the slope of the rate change will be linear.
And as the wind effect tapers off at lower speeds, the effect of Crr will start to predominate, as this effect is also quite linear. One can prove this by calculating the total kinetic energy of the moving vehicle and then start subtracting out the effects of air drag, Crr, and wind.
The kinetic energy of the moving vehicle is dependent upon the square of the velocity, AND the air drag at high speeds is also, as luck would have it, also affected by the square of the velocity. These two effects cancel each other and we are left with a linear rate of deceleration at high speeds. (The Instructables shows this rate changing; i.e. not linear). Once we get to the point were air drag is no longer much of a factor, then Crr starts to take over, and it's rate of change regarding velocity is linear, not a squared function as air drag is.
It gets a little complicated, but I am still collecting data, and will hopefully report my findings later this spring, after I can compensate for barometric and temperature effects.
There is a graph of deceleration over time included below. This graph is in Log/Log format, and the reason which I will not try to explain here, but is quite useful for certain applications.
Hope this helps, Jim.