This is accurate as far as it goes. There are differences in natural performance that disappear with with constraints.

There is no difference with simple material under any conditions. As the material becomes more complex, reading has superior comprehension for natural conditions. This is because readers will speed up or slow down as the material dictates, but listeners don’t usually have or use that control. Readers will also hop back and read difficult passages a second or third time, again something that listeners can’t/don’t.

If you constrain readers by forcing them to read at a fixed pace without review then the differences become minimal again.

So there are definitely differences, but they are tied to the performance characteristics of the medium rather than a cognitive difference.