With-gun robberies are three times as likely as with-knife
robberies to be fatal to the victim[1], and it seems plausible that
this lethality extends to other crimes.

Andy Freeman said:

No, with-gun robberies are not three times as likely as with-knife
robberies to be fatal to the victim. Lambert consistently
“misreports” Zimring’s data.

It is Andy who consistently and wilfully “misreports” Zimring’s data.
Interested readers can look at his November Scientific American
article, his book “Citizen’s Guide to Gun Control”, or the original
journal article (J of Legal Studies 15 (1986):1,16).

A quote from his book: “The death rate from gun robberies is at least
three times as high as the death rate from knife robberies”

And an extract from a table in the book

Percentage Distribution of Robberies and and Robbery Killings by
Location & Weapon
         Commercial          Street              Other
         Robberies           Robberies           Robberies
         Kill-  Non-         Kill-  Non-         Kill-  Non-
         ings   Lethal       ings   Lethal       ings   Lethal
Gun      81     67           67     39           83     43
Knife    13     11           11     15           13     15

Lambert doesn’t like trend analysis;
look at his statistical arguments for the effects of gun control in
Oz and the UK. his stats and analysis are completely consistent
with “the crime rate was decreasing before gun control, but its
introduction stopped that decrease”, which fits the data better than
his model.)

Andy has asserted this before, but has never provided arguments or
analysis to support it. He has not because he cannot. No matter how
many times he asserts it, it will continue to be false. A model was
proposed where there was a decline that ended some years after control
— this was twice as complicated and fit worse than the “constant rate
before, constant rate after” model.

Heck; he doesn’t even bother to control for other factors. He
just asserts that they have no effect,

Perhaps you will show me where I have ever made such as assertion.
Just because you’re into “proof by assertion” doesn’t mean other
people are.

even though crime rates change greatly even when gun control
doesn’t. (Look at the changes in the US murder rate from the 1900s
through the 1990s for many examples, and notice that the low point
corresponds to the least gun control….)

It would be more relevant to look at the NSW homicide rate. After
1920, the rate has been essentially constant, in spite of the
Depression, WW2, a five-fold increase in population, massive
immigration, a doubling of urbanisation, large demographic changes,
and major social change.

It is also interesting to note that the US non-gun homicide rate has
also changed little over that period, so that whatever caused the
change in the overall homicide rate is somehow connected with guns.