Ready for a long, vague but precise answer?
It matters more what the whole package looks like.
As with any engine it's going to depend on what the tube size/lengths are, the cam timing and overlap, and the intended power level and the rpm range looked at.
Common sense applies, headers or manifolds dumping into dual 1-7/8 pipes are choked no matter what on 500 inches.
Manifolds into good exhaust on a mild engine will see the most improvements at the extreme ends of the powerband and minimal in the middle.
With more performance this exaggerates some, with the beginning and the end being more extreme.
Most of the time, seeing a +50 or +100hp gain is due to when it's waking up earlier or hanging on longer, kind of a phenomenon of the curve.
While it's easy to math out where in the rpm range they are more effective, it's difficult to accurately predict by how much they will increase. It's quicker to test on the engine, which isn't quick.
They aren't just exhausting a cylinder, they can actually pull on the intake sooner in the cycle...which can make it an important part of the puzzle.
While difficult to have a relevant comparison by replacing a header with a manifold on a well sorted combo, the header could either look really good or bad the same as the wrong intake.
Was that intake worth 100hp? No, just the wrong one for that combo.
Even more difficult to measure on the typical hot rod type dyno is the effect on part throttle or mileage.
I personally see very large increases at part throttle with Cadillacs, but they are fairly well sorted out.