Why does a coin falls faster when it's flipping as well?

From my experiments with measuring how fast a coin falls, I have consistently measured a faster falling rate for a coin that flips as it falls.

As an example, a coin dropping on its edge from height of $45 \:\rm{cm}$ hits the ground $20 \:\rm{ms}$ later than a flipping coin falling from the same height.

Now here's the catch: I use a microphone to mark the events. I drop the coin off the edge of a table letting it slightly brush off it. The bang noise of this event combined with the noise the coin makes as it hits the hard ground let's me measure the fall duration accurately (I hope). I also take into account the time it takes the sound of coin hitting the ground to come back up to the mic.

Using $\approx340\:\rm{m/s}$ for speed of sound and $9.806\:\rm{m/s^2}$ for acceleration due to gravity, my measurement of height is dead accurate, BUT only for a coin dropped on its edge. A flipping coin constantly gives me a measurement less than correct value.

First I suspected the air resistance, but if that was the case, shouldn't the coin falling on its edge fall quicker?

Any ideas?

-
1) How are you flipping the coin?? Note that most processes that flip a coin will impart some extra translational energy to it as well. Also, if the coin has not yet begun flipping at the time it brushes the mic, then there will be a time discrepancy. – Manishearth Apr 5 '12 at 7:18
Thanks for replacing my numbers with proper ones (I'm not fluent in TeX). Well I drop the coin from about half a cm above the edge of the table while holding it horizontally. When it hits the edge, it start flipping as well as making the noise for mic to pick up. – Mansour Apr 5 '12 at 7:26
Can you comment a bit on the accuracy/reproducability of your set-up? Is this 20ms based on the average of 100 realizations with a standard deviation of 100 ms for example, or is $\sigma$ 0.1 ms. – Bernhard Apr 5 '12 at 7:39
I must warn you that I'm not a physics student, so my method may not be proper. I consistently (100% of time) get a faster falling rate for the flipping coin. The difference however changes but for the 45cm fall, I recorded it between 15ms to 22ms after 20 tries. – Mansour Apr 5 '12 at 8:01
To extend on my remark about the difference in the fall distance of the center of mass: A difference of only 1.1 millimeters in the fall distance would change the fall time by 15 milliseconds. That is on the order of the thickness of a typical small coin. – Jerilyn Franz Apr 6 '12 at 14:38