We don't have any black holes conveniently to hand (the nearest is 3000 light years away) so we've never studied a black hole up close and personal. And, as you say, if we did have a black hole next door we could only study the region outside the event horizon. So black holes are currently theoretical objects - that theory being general relativity.
However we know GR works well where we can test it, and we have no reason to suppose it stops working at an event horizon (not strictly true - see below) so we assume that GR correctly describes a black hole both outside and inside the event horizon.
And what GR tells us is that once any object is inside the event horizon there is no way for that object to resist falling into the singularity, and indeed for all but supermassive black holes that fall takes only a few milliseconds as measured by the falling observer. This is because orbits inside the event horizon would require an orbital velocity faster than light, which of course isn't possible. This is why we believe all the mass in a blak hole is concentrated at the singularity.
But not everyone agrees that GR works at or inside the event horizon. Mostly the objections come from quantum gravity, and the most extreme of them is probably the firewall idea. If this is true the spacetime inside the event horizon may not exist at all in any physically meaningful way.