The claim that a "good guy with a gun" is the only thing that stops a "bad guy with a gun" seems to be rejected by most experts.
It is reported that out of at least 433 active shooter attacks in the US between 2000 and 2021, only 22 (i.e. 5%) were ended by an armed bystander shooting the attacker. Of these 22, 10 were off-duty police officers or security guards, so only 12 were stopped by the traditional conception of the "good guy with a gun" (i.e. a civilian who happened to be going about his day who heroically shot the bad guys).
One objection to this statistic is that we cannot count gun-free zones in the statistic.
So I went back to the data to try to exclude them.
The original data set comes from the Advanced Law Enforcement Rapid Response Training Center at Texas State University, and the Active Attack Data excludes domestic shootings and gang-related attacks.
From 2000-2021, there were 467 incidents listed, so the dataset seems a bit different. It's been over 3 years since the article was published, so presumably some missed attacks got added (none were due to an armed bystander shooting the attacker).
What sort of places are gun-free zones?
The U.S. Concealed Carry Association (which is pro-gun) lists K–12 schools and school zones, Colleges and universities, Federal buildings, Private businesses with posted signage, Hospitals, places of worship and entertainment venues and Public transportation (by which they refer to most parts of airports).
Yet, almost none of these prohibitions are absolute. For example, college and university gun-free zones depend on the state, and even though public, private and religious schools from elementary to high schools are regulated by the Gun-Free School Zones Act of 1990, there're exceptions including being a licensed concealed carrier (depending on state). And naturally, private businesses set their own rules (which may not even be legally binding).
Regardless, let us be as conservative as possible and exclude all locations which might be gun-free zones: factories/warehouses, offices, retail, school and other. We are thus left with outdoor shootings.
Out of 95 attacks that were outdoors, their resolutions were as follow:
Fled - 27
Shot by citizen - 2
Shot by off duty officer - 2
Shot by police - 29
Subdued by citizens - 6
Subdued by police - 14
Suicide after police arrival - 5
Suicide before police arrival - 6
Surrendered - 4
So only 2/95 (2%) of outdoor active shooter attacks were ended by a "good guy with a gun".
Outside of possible gun free zones, the "good guys with guns" look even more impotent, tying for the lowest category of resolution. Certainly, it is categorically, unequivocally false that "the only thing that stops a bad guy with a gun, is a good guy with a gun"; we can see that 3 times as many outdoor attacks ended with the attacker being subdued by citizens as shot by citizens.
Since there were only 6 such cases, I could check their resolution:
2011 January - Arizona, Tuscon - Tackled to the ground
2017 December - Baltimore, Maryland - Girlfriend pulled him out of the car
2018 April - Indianopolis, Indiana - I couldn't find anything on this, nor could Gemini
2020 May - Ft. Leavenworth, Kansas - Hit by a car
2021 April - San Diego, California - Tackled
2021 August - Wrens, Georgia & Graniteville, South Carolina (he was 33, not 30 as in the dataset) - Restrained
So it looks like what is even more effective than "a good guy with a gun" is "unarmed good guys with their bodies".
Of course, pro-gun people are going to claim that there're so few instances of "good guys with guns" stopping gun attacks only because so few people carry guns, and/or that there're so many restrictions on guns, and that if guns were more common, there would be more "good guys with guns" stepping up.
Let us leave aside the point that even if we did indeed get more "good guys with guns" with laxer gun laws, this wouldn't be a good thing since there would be more gun attacks (there is a general correlation between gun-friendliness and firearm homicide rates, even if the effect is not that strong).
So let us restrict our analysis to more gun-friendly states to see if the "good guy with a gun" effect is stronger there.
Looking at the 10 states (i.e. the top quintile) that Ammo.com rates as most gun-friendly during the same period of time, 2000-2021, we get 67 attacks in total, and of these only 5 were stopped by a citizen shooting the attacker. 7.5% is higher than 5%, but still not that great, and it still is very far from supporting the claim that "the only thing that stops a bad guy with a gun, is a good guy with a gun".
If we restrict the analysis to outdoor attacks, as per the main analysis for all states, a grand total of 0 cases (i.e. 0%) were stopped by civilians shooting the attacker.
So in conclusion, we can be even more sure that "the good guy with a gun" is basically a myth and ineffective to stop active shooter attacks.
Of course, one can come up with further copes, for example one is that cases where the perpetrator flees could be because they were shot at by "good guys with guns", but besides being a very tedious process to code all of these cases, there won't be enough information in news reports to rule out this possibility.
In any event, the hypothesis that shooters are cowards and flee because of "good guys with guns" is not supported by the data.
From the original article, we can see that in 184 cases which ended after the police arrived, the attacker only surrendered 15 times (8.2% of cases).
If shooters are really cowards and are so easily scared by guns, you would expect most of them to surrender when the police arrive (where we can be confident that 100% or close to 100% of the police have guns).
In contrast, before the police arrive, we can be confident that much less than 100% of people around the shooter have guns.
Anyhow, even more involved copes can be devised (one I've seen is that just showing a gun to someone will scare him into not committing a crime, so "good guys with guns" can deter even without firing their guns). But people who are emotionally vested in their beliefs can't be convinced anyway, since if nothing else they can and do fall back on making moral arguments about the right to self-defence (so the song and dance about the efficacy of the "good guy with a gun" is really just a smokescreen).
