Tailgator, that assume a normal or at least symmetrical distribution. That's an empirical question and we really don't have a good handle on shots fired in DGUs. However, your point is well taken and correct (except for a stat guy quibble
).
Neal - I'm sorry but you show ignorance of statistical decision making. There is no necessity to go the extremes of Tactical Timmy. As said quite a few times, one decides not to act at a measure of central tendency but for a reasonable cut off in the tail of incident intensity. From good ol' null hypothesis testing, you can decided your statistical power and Type I error cut offs (.01, ,05, etc.)
Most experts in civilian usage have discussed this and a reasonable cut off is that a quality semi pistol and one or two mags is not that hard to carry and probably gives you enough rounds for the 'typical' intense incident. At Orlando or Mumbai - a Glock (or similar) and a mag or two might get your through the day. At the Kenya mall, the handgun armed civilians realized that they didn't have an extra mag and needed one. The police officer, off duty, who intervened in a mall shooting (duh-mind fails - someone fill in) had only the rounds in his gun and regretted that.
It is silly to think that one cannot make reasonable decision levels of risk without going off the deep end.
Remember, modal figures of DGUs say you need no ammo.
If you want a bug - why not. Three muggers and your J frame - go for it.