From what I'm able to gather just in internet research without going and looking at anything...
"THIS LOOKS LIKE A JOB FOR...(you know who)."
- Duck Dodgers
I went into my 1998-2013 homebuilt accident database and extracted the Subaru accidents. I included Eggenfellner, NSI, and Stratus engines. Considering the usual types of airplanes involved, I extracted the Lycoming O-320 accidents as a control group. I did not include any of the aftermarket engine cloners or anything listed as a modified engine.
There were 172 total Subaru accidents, and 453 involving O-320s. However, over 40% of the Subaru accidents (43%) were on rotary-winged aircraft, while only 3% of the O-320s.
Gyros have an advantage in an engine failure; they're basically already configured for an emergency descent. Engine failures are thus less likely to end up as recorded accidents. So I weeded out the rotorcraft and looked solely at fixed-wing accidents.
That left 97 Subaru-powered homebuilts and 439 O-320-powered planes.
To me, the first statistic is one of the most important ones: What percentage of the accidents began with an engine failure. The failure may not be the engine's fault; it could be fuel system problems or the pilot's running out of gas. But when you're comparing engine types, one can assume the portion of pilot/system-related accidents were similar. For a point of reference, 15.2% of Cessna 172 accidents and 18.2% of Piper Archer accidents begin with the engine stopping. I haven't done that many production-type airplanes, but the worst so far has been the Cessna 210: 31.6% (one out of eight 210 accidents is due to fuel management).
For the overall fixed-wing homebuilt fleet, 33% of the accidents started with an engine failure.
For the O-320 powered fixed-wing aircraft, it's 29.5%. Better than overall.
The Subaru group?
Over half: 52.6%.
Looking at the causes (percentages are of total number of accidents):
Engine internal issues (rods, bearings, etc.): 5.2% Subaru, 1.4% O-320.
Ignition/Engine Controller: 9.3% Subaru, 1.4% O-320.
Reduction Drive: 3.1% Subaru, none on the O-320s (remember, this is fixed-wing only).
Cooling System: 6.2% Subaru, none O-320.
Fuel system in the engine compartment: 3.1% Subaru, 0.9% O-320.
Almost ten percent of Subaru-powered fixed-wing homebuilt accidents are due to systems the O-320 doesn't need... Propeller Speed Reduction Units and liquid-cooling systems.
Despite there being more than four times as many O-320s in the accident lists than Subarus, the Subaru group showed more cases of ignition system/engine controller issues and practically as many internal engine failure cases.
Ron Wanttaja