I've had several conversations in the past few weeks on the different modes of dissemination and the REF's undervaluation of the book collection. The argument goes that essays in a book collection are less valued because they're not peer reviewed. Except... they are. They're just not shipped out by an editor of a journal to multiple experts to formulate consensus.
In some ways, the edited collection is reviewed to a greater extent. If we accept that academic capital is accrued on the value of reputation, an academic who edits such a collection is staking his or her reputation (an investment of academic capital) in exposing their name. The same cannot be said of reviews for a journal. I may be discredited from reviewing by a cursing editor if I gave a favourable review to a piece that was flawed (I wouldn't do this by the way, this is hypothetical), but I would not be exposed. The loss of academic capital that I would incur from such behaviour is minimal compared to editing a collection.
It's probably for this reason that, when researching within a sub-field of literature, I often find that edited collections contain much more material of interest. Pieces in these volumes have the freedom to specialise in a way that journal articles in the "top-flight" publications often cannot. In short: trust the editor, trust the material is one of my credos and I think it is one of many flaws that the REF is so superficial in its non-acknowledgement of models of academic capital, focusing on, and fostering, exclusivity and competition. In fact, much of my thinking (when I consider which publications I most value) leads me to conclude that I would far rather that most of the funds invested in journal subscriptions (and -- coming soon to a university near you! -- open access fees (ugh)) in the humanities were paid to publishers of monographs and edited collections who actually invest time and effort in their authors. In this way, publisher value would be made apparent.