Dan Day wrote:

I’ve reread that a number of times and still can’t figure out
exactly what Kellermann is trying to say. The best I can make out
is that Kellermann is claiming that self-defense homicides are not
legally justifiable (?)

You missed the bit where he defined the meaning of those terms:
“Self-protection homicides were considered “justifiable” if they
involved the killing of a felon during the commision of a crime; they
were considered “self-defense” if that was the determination of the
investigating police department anf the King County prosecutor’s
office.[11]” Reference 11 is the FBI definition of “justifiable
homicide”.

No, I saw that (upper left on page 1558 of the NEJM publication).
I didn’t see as how that clarified Kellermann’s choice of wording
on page 1559. And it still leaves open the following questions:

(1) Why the distinction between “self-protection homicides” which
are “considered ‘justifiable'”, versus “self-defense”? Ask five
different people and you’re likely to get five different opinions
on which homicides fall into which category.

Because the FBI and the police make the distinction. The only national
data available is for “justifiable homicides” and you can’t refer to
that without making the distinction.

(2) Why are only “self-defense” homicides which the prosecutor’s
office
determined to be self-defense counted as self-defense?
Why not the court’s decision?

Because criminal courts are biased towards making sure that people are
not punished for crimes they do not commit. So, someone who was
probably guilty of criminal homicide, but not guilty beyond reasonable
doubt could be found not guilty with a defence of self-defence.
That’s fine, but if we are trying to get the best possible estimate of
the number of self-defence homicides we should use the standard of
civil courts — “more probable” rather than “beyond reasonable doubt”.
The decision of the prosecuter’s office seems are reasonable way to
estimate whether someone is more likely to be guilty or not.

(3) Why are “all homicides resulting in criminal charges” considered
“criminal homicides”, when surely many of them were later found to
be justifiable by the courts? (Even many clear-cut cases of
defense are referred to the courts by prosecutors because it’s the
court’s job to sort out such things, not the prosecutor’s job).

(4) Getting back to the core of the original issue, why did Kellermann
choose to word his passage on page 1559 as “Less than 2 percent of
homicides nationally are considered legally justifiable”? Most
readers, it seems to me, would take this to mean that 98+% of
homicides are not considered “legally justifiable” — that is,
they’re illegal, or they’re unjustifiable, or both.

I’m sorry, but I see no other way to refer to justifiable homicides, as
defined by the FBI, other than to use the term used by the FBI. Not
only does Kellermann give the FBI definition (on page 1558), on each of
the three occasion when he uses the term, he provides a cite to the FBI
definition (cite 11). Furthermore, underneath the passage in question
you can see a table classifying the deaths and it is quite clear that
the 2 justifiable homicides in the study are a subset of the 9
self-protection homicides.

If, after all this, a reader still manages to misunderstand, I don’t
think it is Kellermann’s fault.

And, in any case, this is still no excuse for Kleck’s misrepresentation
— Kleck, at least, is well aware of what the FBI mean by “justifiable
homicides”.

The final sentence in that paragraph (“A majority of these
homicide victims were residents of the house or apartment in which
the shooting occurred”) throws further confusion into the mix,
seeming to imply (by juxtaposition with the preceding) that if you’re
a “resident of the house”, you can’t be the result of a self-defense
homicide.

I find your reading of that sentence truly bizarre. Why not take it to
mean what it says, and not something else?