The statistical approach whilst attractive is fundamentally flawed. Just because an event happens frequently and even more frequently than other events does not mean that it will prove to be the cause of the next incident (if you want a spectacular illustration of this take a look at the Fukushima nuclear power station). The fire explanation is just as far fetched as any of the other explanations - perhaps even more so as it requires an even greater number of unlikely events to happen (which does not preclude it from being the explanation). However, even though the rarity of the hijack or pilot deviance explanation is significant this particular line of investigation still fits the known facts better.