Why could a B+ tree be less efficient than a B tree in random search?

Question

I was wondering why a B+ tree could be more inefficient than a B tree in a random search (I read this in a theory book that doesn't explain why) considering that the only difference - far as I know - between these two types of data structure is that B+ trees got in every leave node a pointer to the next node. I think there is something that I am missing.

Could someone explain me further differences between these two data structures?