Yep, while hiding and calling the cops is generally safer during a robbery, the attitude that causes people to defend their property is the same one that results in gun ownership. Add in people being shot with their own weapons and the home protection benefit becomes very complex. There is an idea that gun ownership deters criminals but at the same time it promotes criminals buying guns escalating the situation.
PS: For a somewhat less contentious example, video games have been linked with a significant drop in crime. However, because that does not agree with their biases people focus on the direct effects to support their argument.
I think you'll find there tends to be a correlation between using [protection against bad thing] and being at risk for [bad thing]. As an example, I'm fairly certain you'll find a correlation between wearing a helmet while driving a car and being injured in car crashes because most people who wear helmets while driving are racing or doing stunts. I also suspect people with burglar alarms lose more money to burglary than people without if no other variables are taken in to account.
I don't think an unbiased study has been done that accurately identifies causes looking at gun ownership and crime. My suspicion is that it would be possible to do, as I described above.
Add in people being shot with their own weapons...
This is a pretty minor effect, unless you are including suicides. In 2007, there were only 613 accidental firearm deaths. For comparison, there were 1039 "transport, other" deaths (I believe this means plane/boat crashes).
All of these numbers are fairly low. Adusting for the percentage of gun owners and you end up with a higher rate than just 5% of all deaths from firearms being accidental. And when you start looking at low risk populations (for violent crime), accidental death from gun ownership is fairly significant relative risk.
PS: For a somewhat less contentious example, video games have been linked with a significant drop in crime. However, because that does not agree with their biases people focus on the direct effects to support their argument.