I couldn't agree with you more - I also find this very strange indeed! I don't think it would be cumbersome to read with numerical citations linked to a references section. It would give me a bit more confidence in the rigour being applied at the investigator end.
For example, while I find the AAIB output very good at determining technical aspects (i.e., which bolt was fractured pre-impact), I'm often left with the distinct impression that opinion is being presented as fact when it comes to suggesting aircrew behaviour - while accepting it's a difficult area to establish.