I haven’t finished reading the study yet, but some thoughts so far:

• Those that believe they’re at greater risk of being shot (for example, if they live in a high-crime area) are more likely to want to possess a gun.
• Those that had a gun but had “no chance to resist” (basically, they were uninvolved and caught a stray bullet) had a 4.46x higher chance of being shot than those not in possession. This seems to indicate that the increase is almost entirely correlation, and not caused by possession. Stray bullets can’t tell if someone is in possession of a gun or not.
• Those that had a gun and had “at least some chance to resist” only had a 5.45x higher chance of being shot than those not in possession. This is only 1.22x higher than the “no chance to resist” group. Pretty impressive, considering that this includes scenarios where the attacker got the drop on the victim, as well as “mutual combat situations precipitated by a prior argument” where the fight might have been avoidable through deescalation.
• As far as I can tell so far, the control group isn’t anything like “unarmed people that got robbed”, it’s random people in Philadelphia that were interviewed about what they were doing at the time their counterpart in the experimental group got shot.
• Engaging in a gunfight with an assailant and being shot may be preferable to cooperating in the hopes of being left unharmed, depending on the situation. Particularly when the assailant is a violent bigot that wants to kill (and maybe torture) you anyway.