Update from comments: Here is same data with less smoothing to calculate derivatives The peak acceleration during the first bounce is closer to 100 m/s/s where as before it was 60 m/s/s. The motion detector has max data collection rate of 30 Hz.
Bouncy Ball
Teach. Brian. Teach. 
Tracker with video would give crisper acceleration spikes. The software you are using is really smoothing them out. It may be fixable (if you have access to the source code)—see https://gasstationwithoutpumps.wordpress.com/2011/11/08/tracker-video-analysis-tool-fixes/
This is raw position data from a motion detector sampling at 30 Hz. Logger Pro defaults to taking derivatives by using 7 data points (3 before and 3 after). I can change that setting so it will smooth less.
The Tracker changes I mentioned will do a much better job of matching bouncing-ball data (whether from a position sensor or video), as they do not assume that acceleration changes slowly, but allow the possibility of an acceleration spike between samples, which is the model that is needed to match bouncing-ball data.