The fact that he murdered his wife is indicative of the fact that he is capable of murdering another person in a fit of rage. Certainly it's not true that "the murderer is no more likely than anyone else to commit a crime."
It was the condition of a hypothetical scenario that the study was correct. Obviously. You can discuss what evidence is in the real world, but scenarios are meant to exercise your reasoning powers in the steps beyond what you think the actual facts are.
This is no different from thinking about what would happen if you could teleport. The correct answer isn't "but you can't do that".
Sounds nice, but isn't it a very common motive that the aggrieved want "some sense of fairness" in punishment? You hear it a lot after publicised trials, the relatives will offer their opinion on whether the punishment was appropriate for their loss. Doesn't it seem a little weird that only society's interest is what matters?
There's a rather large can of worms in the ideas you speak of.
Safety, for instance, is not an obvious thing. Suppose a study comes out identifying that 50% of members of a given group (say geographical, skin color, education, etc) will commit a heinous crime (and 50% is way more than normal in the population). There is no further specific information; as far as members of this group, it's a coin toss.
What are we to do? The ones who haven't committed a crime yet are a risk. Can we lock them all up? I doubt it.
Here's another interesting one. A study appears in which it is studied what happens to jealousy-murders, eg where a guy finds his wife with another guy and murders them. Lets say the study concludes that once the murders have happened, the murderer is no more likely than anyone else to commit a crime. It's totally safe to let him out now, because it's safety we care about? Seems wrong. But many countries do recognise this kind of enraged murder as a special circumstance in sentencing.
Anyway, not saying I have the answers. Ethical questions have been vexing for a long time.