As far as I know there’s not a site that tracks that info. As a stats guy, I will say that it’s going to be incredibly difficult to get real, good data about this kind of thing. Of course you have the issue of underreporting, but the larger problem is one of definitions. Let’s look at the question:
is there a website that does the best it can to track when a legal gun owners saved a life or prevented a crime such as muggings, home invasion etc.
How do we define “lives saved”? Is it just the legal gun owner? How about if they’re with their family - do they get counted too? Hostage shots only? How do we count other bystanders? Just those in the immediate area? How do we define immediate area? Conversation distance or range of the firearm used? If the incident is in a crowded area, is it treated as a minor event (just the good guy’s life saved) or do we treat it as a mass shooting stopped by a good guy (lots of lives saved)? If it happens in a building, is everyone in the building counted as a life saved? That could be a very high number if you’re in a commercial building, even if most of those people weren’t actually in any danger.
Not trying to poke holes in your question or anything, but all of these things need to be answered in order to make a dataset.
So the problem becomes what is the motivation of the person/group collecting the data? An anti-gun group would use the most restrictive definition (likely only the law abiding gun owner) and a pro-gun group would likely want to cast the net as wide as possible to get the highest number on “lives saved”.
As an aside I would also like to know stats on negative outcomes, the old argument of being shot by your own weapon.
Again, the problem is one of definitions. Most negative stats are likely going to be gathered by anti-gun groups who have a preference for creating stats that support their position.
It’s been a while since I looked at that particular stat, but if I recall correctly the text was “a gun is X times more likely to be used against you
or someone you know”. So while it sounds like “if you have a gun it’s just going to be taken from you and used against you” it’s actually more about >50% of gun deaths being suicides.
Additionally it also counts if there’s some sort of relationship between the gun owner and the assailant. Which still sounds bad until you see that the yard guy breaking into your house at night counts as “someone you know”.
If your data source is a survey, you also need to look at the wording of the questions. Let’s use the “most gun owners support Universal Background Checks” stat as an example.
The best way to ask the question is to give a summary of what the particular law plans to make illegal, and then ask if the individual supports that initiative. (Though the technically correct way is to ask the question a number of times with other distractor questions preceding it to determine if an individual is sensitive to the framing of the question).
A more biased set of questions would be: “Do you think someone should go through a background check before buying a gun?” If “Yes” they are marked as supporting UBCs. Alternatively you could ask “Do you support a law making it illegal to give a gun to your neighbor without explicit government approval and a fee?” If “No” they are marked as not supporting UBCs. Both of these are arguably true statements about UBCs, but the wording of the question effectively predisposes the population to choose a specific answer.
Anyway, in order to have anything useful, you’d need good/consistent data (difficult to find) and good definitions (given the contentious nature of firearms policy - almost impossible). The above also applies to all statistics, not just gun data.