While designers/bloggers tend to cite scientific studies advocating sans over serif (or the other way around depending on the context), no conclusive scientific evidence to prefer one or the other [1]
However, the historical reason why sans-serif are generally recommended over a serif on a computer screen was a matter of practicality. The computer monitor simply couldn't replicate the tiny detail required by serifs. Higher resolution screens are becoming more common place, which is why you'll see more serifs in use online (IMO... correlation vs causation and all that). Somewhere along the line someone cited scientific articles to help justify this practice, but I'm convinced it became common place was really for practical reasons/aesthetics more than anything.
That's an excellent article you linked -- thanks! From it I gather that no one really knows why serifs exist in the first place; that there is absolutely no meaningful evidence to support a contention in favor of either style on the basis of legibility; and that, at basis, the question is one of aesthetic preference.
Here's an interesting anecdote I just remembered, though. A psychology professor of mine was attempting to replicate another researchers work. The result couldn't be replicated when the stimuli was displayed in sans-serif fonts, but it was replicated when they finally used the font the initial researchers did: a serif.
I'd bet serifs/sans may have some sort or psychological implications but I doubt it's as straightforward a relationship as often presented