Blog readers—and since you’re reading this, that includes you—appear to having been doing something right. At least according to Daniel Lakens, an experimental psychologist who publishes the blog The 20% Statistician. In an interesting post, Lakens argues that blogs have higher scientific quality than journals.
That’s a startling claim, especially for those of us trained in the sciences where journals are considered the gold standard for transmitting scientific knowledge. Lakens offers 5 reasons for his assessment1. The first is basically that blogs more closely approximate the ideal of reproducible research whereas Journals are just beginning to embrace the idea.
He also notes that blogs have better error correction. Generally, if an error is discovered in a blog, the poster will correct the error and note the update in the original posting. Journals can’t do that of course even if they are committed to issuing corrections. Almost always the correction gets lost and unseen by readers.
His other reasons boil down to “gatekeeper” issues. Most journals are hesitant (or refuse) to publish articles that go against the common wisdom—the debate on climate change illustrates this nicely: whatever your position on it, there’s no denying that deniers have a hard time getting published.
Anyone can publish a blog so you avoid issues of eminent researchers being given preference and other biases of editors and reviewers. Everything is, in short, out in the open for the reader to judge. Peer review is a useful device—or at least it seems like it should be—for vetting articles so that readers can presume a measure of accuracy in journal articles but the shockingly high irreproducibility of the results in many journal articles casts doubt that it’s actually working as intended.
Lakens’ post is a provocative and interesting read; it’s well worth a few minutes of your time if you’re involved in scientific publishing or just like contrarian views.
Lakens says up front that he’s mostly familiar with journals in Psychology but his arguments seem generally applicable.