The suggestion is for libraries used only by academic researchers.

(A) is tempting because it seems to offer alternative measures of usefulness, but nonacademic readership is nonexistent and therefore irrelevant in the libraries we’re talking about.

The suggestion involves comparing journals within a given discipline to one another to decide which deserve subscription, so information comparing different disciplines to one another is irrelevant.

Also, for (B) to be relevant, length would need to correlate with frequency of citations, which it doesn’t.

The suggestion is for libraries only used by researchers. For (C) to be relevant, we would need some indication that these libraries’ status as “general public keep out” will change. As it stands, (C) just tells us the general public is probably screwed.

(D) challenges frequency of citations as an accurate measure of usefulness – turns out lots of useful articles go uncited because researchers are snobs.

(E) provides a reason to carry more than one journal, but it doesn’t undermine the accuracy of citations as a measure of usefulness. There’s every reason to believe these multi-journal flame wars will be riddled with citations back and forth, making all the participating journals seem useful according to the suggestion.

It’s pretty reasonable to anticipate that an answer choice will say usefulness in a single discipline isn’t a great decision metric because what about interdisciplinary usefulness? (E) is perhaps designed to scratch that itch – but note that the drama in (E) takes place entirely within a single discipline.