I have carried out a very similar study to Andrew Oswald's but I drew very different conclusions from my results.
I studied the citation history of more than 600 papers published in 1990 in six management science/operational research journals.
The results were clear - papers in top-quality journals received many more citations than those in less prestigious ones.
While the statistical distributions may overlap so that highly cited papers in a low-quality journal can have more citations than lowly cited ones in a high-quality journal, overall the distributions are distinct.
There are also factors other than prestige that determine the number of citations that a journal receives generally, for example width of coverage, type of paper published and whether or not it is American (as US titles have a much larger readership).
What is very difficult to discover is the underlying causality. Do the papers in top journals get cited more because they are, in fact, better quality papers (because of more stringent refereeing, higher standards and people submitting only strong papers) and would thus get highly cited wherever they were published? Is it simply the prestige of the journal that leads people to perceive the paper as better and feel that they have to cite it? Or is it, most likely, a combination of both? It would be hard to design a study to disentangle these two effects.
John Mingers Kent University