Another view could be that the psychopathic behavior may get you fired in a corporate setting and gunned down in a criminal one, so it may be simply that psychopaths last longer in non-violent areas of society.
I guess that would be corrected for in the above-mentioned "The correlates included demographic and status variables" ? Lifespan is objective so rather easy to correct for.
That's assuming you could account for the rate of psychopathy among those criminals who died before they could be tested for it as well as those who died before they were caught as criminals. Lifespan will also probably be different for criminals on the streets vs. inside prison, so I'm not sure you could simply adjust for that.
The original commenter amended his post to say that he mis-remembered about it being higher than in prisons, anyway.
