Even this blog entry is false.

Seed Magazine, the greatest magazine on Earth, has a short summary article about the work of John Ioannidis, most specifically his article in PLoS Medicine entitled "Why Most Published Research Findings are False" . I haven't read the whole article yet, since it's quite long and involved, but it implies some interesting ideas; essentially, that almost all research is actually false because of bias on the part of the researchers, a scientific culture that demands publication, and most importantly, abuse or ignorance of statistics. Since I haven't read the whole thing, it's hard to analyze, but I like the part about misinterpreting p-values. The 0.05 p-value is the gold standard of statistical significance. But Ioannidis proposes a new measurement of significance that more directly measures the probability that the findings are true. The new measure, PPV, is similar to p-value in that it depends on alpha and beta (the probability of type I and II error, respectively), but also the probability before the study that the relationship exists, as well as the bias from the researcher in the form of selective reporting and distortion of data.

It's an interesting report, but it's hard not to ask the question of whether or not it is true itself, given its own findings. It seems a little Godelian to me; it's like saying "this statement is almost certainly false." Maybe he stumbled upon a new proof of incompleteness, or at least a real-world application!