Not making them so boldly, and instead considering the possibility you might be wrong.
And I am. I am considering the possibility that I might be wrong.
So you accept the possibility that war might exist and your understanding of the principle of rationality is wrong?
Which are very often contradictory both to each other and to known facts. Anti-vaxxers have experiments supposedly showing that mRNA vaccines cause subclinical myocarditis in around 3% of the population, way more often than COVID itself does. Neal Barnard and some other people have experiments supposedly showing that some weird diet is better than insulin for managing type-2 diabetes. Trusting experiments whose results go against the basic principles of science doesn't get you very far.
No, it is usually contradictory to each other and known facts.
They might appear contradictory.
Especially when you don't understand the possibility of random chance giving the result, which is why there is a p value in the first place, and why a p value of 0.05 isn't all that great.
That means if you conduct 100 experiments to test something, 5 of them will give the result just by chance.
It also means if you want to look for a positive link, if you test 20 of them, you would expect 1 to give a result by random chance.
That isn't a result because of an actual causal link. I wouldn't even call it a real correlation. It is just looking at the noise.
This was even demonstrated by one study which intentionally looked at lots of different variables so they expected to get one by random chance:
https://gizmodo.com/i-fooled-millions-into-thinking-chocolate-helps-weight-1707251800.
With such issues of noise and randomness, a single experiment is not enough to reach a strong conclusion of causal relationships for biological sciences. Especially when you have a very small sample size.
That is why there are meta-studies, which look at multiple studies.
That is also why there are placebos and blind studies or double blind studies.
Another big issue is understanding the difference between correlation and causation.
You can have a study that shows a correlation, but that correlation doesn't necessarily mean there is a causal link directly between the 2.
It could just be a chance occurrence, or it could be both are caused by something else.
One simple example of that I saw recently was the link between sunburn and ice cream consumption.
It isn't that one caused the other, it was both being caused by it being a hot day with sun out making people more likely to go to the beach and get sunburnt, as well as being more likely to eat ice cream.
But people not understanding that and seeing the result might say eating ice cream gives you sunburn.
A proper understanding of uncertainty and random noise and the like means you can get studies which produce apparently contradictory results.
If you would like a more physics based one, it would be like one thermometer saying it is 23.0 degrees, while another says 23.1. They are different numbers, and might appear to be contradictory, but within uncertainty they are the same.
As for "known facts", there is the issue of if it is a fact, or merely a belief.
I am yet to see a known fact be contradicted. But I have seen plenty of beliefs be contradicted.
There also experiments you can do to show guns can be fired.
Like...?
The simplest one, go to a shooting range, pick up a gun which has been loaded with a standard round and fire it.
If you would like more variables, then it depends upon what you have.
You could try a classic flintlock style gun, start by just loading the bullet with no gunpowder and see if it fires.
Then try varying amounts of gunpowder until you reach the normal load.
Alternatively, you can try a modern firearm, with rounds loaded with varying amounts of gunpowder, one with just primer, and one with nothing inside.
Ideally you want it placed in a fixed stand for a repeatable firing position, with a high speed camera watching the muzzle to see if the bullet comes out.
If you had access to the appropriate facilities (i.e. a blast range) you can even do the same with a bomb, by making bombs with different amounts of explosives inside including a control with no explosives.