We're talking probabilities here. It's more probable that a 5V board will fail than a 3V board, probably for several reasons.
It seems the 5V boards were running closer to the maximums as far as power is concerned. Many times one or more diodes fail inside the board sandwich, and since they are not accessible, it's not possible to attempt a repair.
5V boards consume more power and run hotter than 3V boards, thus shortening the MTBF of many components.
The 5V boards are older, and the electrolytic caps, which are a significant failure possibility, are older.
I've personally seen that the 330uf cap of the WinCap brand fails more often than others on the board. It's my contention that these may be some of the defective capacitors that were shipped around 2000-2002 from various Taiwanese makers. These were used in various products for a significant amount of time after manufacture, so failures may have extended much later. See Capacitor Plague for more information.
All electrolytic caps will eventually fail, their failure rate is based on the initial specification and the operating temperature. The operating temperature inside many locomotives if fairly high, resulting in a shorter lifespan.
Even unused electrolytic caps can dry out. Since a number of failures are on the very first power-up of a locomotive long stored away, it's hard for me to blame heat for the failures, they don't get a chance to get hot. Most semiconductors will last longer than we will when stored in normal temperatures. So, it seems more logical to me that a defective cap that could have dried out in storage would be more likely to fail than a semiconductor that should have a practically infinite shelf life.