It's one thing to skim an article and write a blog post taking off from the article or commenting on an aspect of it; blogs are not journals. It's quite another thing to submit a paper to a refereed journal and cite an article for an opposite argument than the one it actually makes, or for some proposition it has nothing to say about. C. Blattman draws attention to the editors' notes in the current APSR which mention this as a widespread problem. If it is a widespread problem -- and although I don't follow the journals all that closely, I'm inclined to think it probably is -- it's deplorable. There are two possible reasons for this: carelessness or outright dishonesty. I've also noticed problems with the way books are cited, which the APSR editors also mention. Recently I've seen an article that cited a book for a proposition that the book had little or nothing to do with; I've also seen an article, in a 'top journal', that cited books this way, not once or twice but quite a lot: Smith 1998 or Jones 2002 -- no page numbers, no chapter numbers even, just the author's last name and the publication date. This is often both unhelpful and potentially misleading, as it can create the impression that the entire book fully supports whatever point in the text it's being cited for. It's certainly not a completely indefensible practice and there are times when it's entirely appropriate but, IMO, it should be used with some discrimination.
Just so I don't come off as a disgruntled grouch, there are of course still articles published which cite works in an exemplary way. But this shouldn't really be an issue at all. Part of the problem is that there is too much being published and authors are afraid their submissions will be rejected for failure to cite some allegedly relevant piece, so they will cram in tons of citations whether they have actually read them carefully or not.
I have had only one experience with submitting an article and getting readers' reviews of it -- it was about five years ago and the article was, admittedly, not fully baked (it was not drawn from my dissertation but was about something entirely different). There were two reports, one quite cursory, the other impressively long and detailed, both recommending rejection. The long, detailed report faulted the piece on several grounds ("undertheorized and empirically weak" is the phrase that I recall) but one of the criticisms was that I hadn't cited this, that and the other. The reviewer was right: I hadn't cited X and Y. Should I have? Quite possibly. On the other hand, I would rather have had the article rejected, as it was (I never did anything with it after that or tried to get it published anywhere else) than have cited works I hadn't actually read or at least looked at with sufficient care to determine what they were actually saying.
I think this post probably should be filed under "rants" but I'm just going to put it under "miscellaneous."
(P.s. I've published book reviews but they're a different kettle of fish than articles.)