The journal Social Psychology has issued a call for proposals for a special issue of their journal to be published in 2014. The name of the issue will be “Replications of Important Results in Social Psychology” and the call is available here.
The main reason that this is necessary is that replication isn’t “sexy”. Journals are interested in novelty and interesting findings, and so when they are offered a choice between printing the same thing twice or printing something new, they will always choose the latter. However, this means that those important replications are never formally published, and this can be significant when the replications cast doubt on the findings of the original study. Instead, we hold to the findings of a small number of groundbreaking studies which become effectively immune from high-profile critique due to a form of the “file drawer effect” (which I have mentioned before in the context of clinical trials). Such is not the way of science, so kudos to the editors of Social Psychology for attempting to turn the tide on this important issue.
Katatrepsis 
Do you think a journal(s) should be started that would deal solely with replicated studies? I know it wouldn’t be the most popular journal, but the studies should get cited in papers that cite the original experiment, and it would help tease out experiments that aren’t consistently repeatable or maybe fluke experiments with P-values that happened to land on “the right side”. Plus it would allow researchers, who have to repeat experiments for whatever reason, to get more publications.
I think the problem is the obsession with only publishing those studies that will get cited. What we need (and I was chatting about this with some staff at Leeds last week) is for the Research Councils (NSERC in Canada, NERC in the UK) to publish all of the science. That means it is only ever in the public sphere. The result is a big melting pot of publications that can be sifted through and ranked by the interest that they generate among the readers. This is how PLoS ONE works – 5000 papers per year and let the audience sort the wheat from the chaff. Only once we have got rid of this “impact factor” nonsense will we be able to push forward.
That’s a really good idea! How would peer review work under that model? Would it be possible for every paper to be peer-reviewed before publication? Also, should the authors be charged for submitting the articles, or should it be funded with taxes? I’m just curious what you think and may have discussed with Leeds’ staff.
Peer review would be the same as usual: a small number (2-4) experts would be invited to review. The idea would be to draw traffic away from major publishers, so it wouldn’t be an increase in reviewers but a reallocation of their time. The financial side should be a part of the funding structure, but economies of scale (and a lack of profiteering) should keep costs down. I have heard editors say that a good open access journal could be run on a contribution of under $100 from the authors, as opposed to over $1000.