The trouble with the statistics is that they are all encompassing.
Yes, the stats show that a single engine failure in a twin is more likely to result in death or serious injury than an engine failure in a twin. Obviously a twin is twice as likely to suffer an engine failure than a twin.
But this encompasses lots of different pilots. Many many of whom will probably be barely able to afford to fly it, and as a result not be very current. On top of the low currency, they will be even less current on asymmetric flight. It stands to reason that if they don't get to fly it very often, then they will want to use it to go places etc when they do get to use it. They won't want to 'waste' their time practising.
A twin owner on the other hand is probably more current in general and more current on asymmetric flight, and therefore their risk profile is probably much better than the general statistics.
Of course this is total guess work on my behalf, and I could be totally wrong.
