There were 3485 shootings on police record for the duration of the study so it's a very large percentage. But this does not matter, for statistical purposes it is waaaayyyy more than necessary.
Something to consider - a larger sample size is more likely to give you statistical significance regardless of effect size. So, the more data you look at, the more likely you are to find statistical significance regardless of true effect size. A bigger sample would produce higher significance.
Not from the New Scientist report, the article tells us quite a lot about the observable characteristics of the matched groups.
Except for the notable fact that the crime was committed with a gun, and not with a hammer or bare hands.
If it makes you feel any better, with very few exceptions, every victim of a bare-handed attack also had bare hands.
I don't think there is anything mystical about guns changing the behaviour of their owner, unless they were completely unaware of their gun possession.