We can test that concept -- for example, comparing rates of serious violent crime in the US (high rates of gun ownership) and Japan (low rates), or comparing such rates between states with similar levels of urbanization but different gun ownership rates (e.g., WV v. ME, FL v. MA, TX v. WA, etc.) If having a higher probability of a potential victim being armed deters crime, we'd expect the higher-gun-ownership places to be safer. Is that what we see, here in the real world? Or does that concept fail to survive the journey from conservative hypothesis to actual reality?
Alternately, we could compare states that start out at a similar place, in terms of violent crime, but where one then has changes that encourage more people to carry guns (e.g., easing of concealed carry requirements), and see what happens to their violent crime rates relative to the states that didn't have such a move. If a rising apprehension that a potential victim might be armed actually deterred crime, we'd expect to see a pattern of places moving to more widespread carrying of guns experiencing better violent crime trends than those that didn't. Is that what we see?
That question has been studied:
https://www.theatlantic.com/politics/archive/2017/06/right-to-carry-gun-violence/531297/
I think the test of a good theory isn't whether you can make it sound plausible, nor whether it appeals to your preconceptions, but rather whether it lines up with the data.