On the other hand, the odds of the majority of passengers surviving a controlled-flight water landing are pretty good, while the same numbers for an urban landing are not so great. So "worst case" for a ditch is a significant number of survivors.
That might very well be true in 70F water in the summer time. However if the plane here had broken up and caught fire on impact I highly doubt if very many (or indeed possibly any) passengers would have survived. Those not immediately carried to the bottom or burnt to death would probably have drowned wthin minutes in the icy waters. And passenger boats would not have been nearly so keen to rush to the rescue if the river had been ablaze with jet fuel.
The sad example of Air Florida Flight 90 would also rather suggest that ditchings in to freezing rivers in January do not normally go at all well for passengers or crew and that despite his clearly exceptional level of skill as a glider pilot Captain Sully cannot have set down on the Hudson at all confident of the kind of result he actually achieved. That result was no doubt his best possible case but probably only a 5% or so expected probability at best.
Any review of the history of loss of all engine power incidents would probably have suggested that some kind of controlled landing on to a runway was far more likely to have more survivors so one can only assume that the pilot took his decision virtually certain that an attempt to return to La Guardia would not result in a landing on the runway. And British Midland at Kegworth clearly indicates just how appalling the results usually are when loss of all engine power and an attempt at a controlled landing instead sees the afflicted aircraft landing just short of the target runway.