Herd of p-values spotted approaching significance
A group of field biologists in Tanzania have observed a herd of p-values approaching significance. It is the first recorded sighting of this mysterious behavior, that has previously only been speculated to occur in the wild. Lead investigator Bruce Rosen is still excited about the sighting. “It was amazing! The α-male, a majestic 0.06, was seen slowly but surely approaching significance, followed closely by a small group of marginal p-values. First, the p-values formed a uniform formation, possibly as a distraction. But shortly after that, the herd slowly but steadily approached significance.”
Blood samples of the elusive creatures shed light on possible genetic mechanisms underlying the behavior. Rosen explains a surprising finding: “Some of the herd had a deleterious mutation causing them to have two tails. Interestingly, these specimens were twice as slow in approaching significance as the specimens with one tail.”
Rosen is hoping to extend his expeditions in the near future. “This is just the beginning. After seeing p-values approaching significance, what we really want to observe is p-values retreating from significance. But that kind of behavior as never been reported, even by the natives.”
Sadly, it is believed that only about 1 in every 20 wild p-values ever reach significance. It’s a harsh life in the statistical jungle.
Keep snooping. There are more out there, Neuroskeptic!
I thought that wild p-values just below 0.05 were more common than the one just above 0.05 (http://www.tandfonline.com/doi/abs/10.1080/17470218.2012.711335). Maybe not in Tanzania…
It is sad but true: p-values above 0.05 are heavily neglected by scientits. This is particularly shocking, because these make up ~95 % of the whole p-value paopulation. I applaud the authors for their effort to make a first step towards ending this discrimination. Stop p-value racism now!
Loving this blog! Mind if I pinch your picture (and link back)? It would be perfect for a little post I’ve got planned!
@neurobonkers go ahead!
Thanks! 🙂
Is there a plural noun for p-values? a “striking enrichment” perhaps?
of course some of us love to find these beasts, they can be tricky to find when you want them. but i’m just a misfit, perhaps
And some people believe that the more .05ers you think you see, the less you should believe that any are actually there at all.
Actually, I had a p-value the other day who was retreating from significance. It was from Levene’s test, really close, but fortunately it was retreating!
There are legends of retreating p-values told in the oral tradition. But these tales are unfortunately rarely found in written form.
My p-value was approaching 0.05 but was quickly made redundant as it was swallowed up by a beta greater than 0.70; the smaller alpha got the bigger beta became.
I really believe this particular blog post , “Herd of p-values spotted approaching significance Collectively Unconscious”, rather entertaining
and the blog post was a fantastic read. Thanks for your effort-Demetrius
This piece of writing is in fact a pleasant one it assists new web viewers, who are
wishing in favor of blogging.
The biologists studying this phenomenon need to take account of evolutionary biology. As the diagram makes clear, the problem with the Tanzanian herd is not having one or two tails, but rather the size of their heads, which makes it hard for their tail p-values to push them towards significance, just like top-heavy sperm who end up failing the reproductive grade. Neither are their angular head shapes conducive to navigating the difficult vaginal channel towards significance and publication, but perhaps these p-values are an ancient proto-family from the Olduvai gorge.
Modern p-value genera from N. America and Europe have certainly been aided by the efforts of CV-driven researchers who use culture mediums consisting of over-extensive data screening, cherry-picking from large multivariate problems, and the occasional falsification of data, because “we know we’re right.” Some small portion of the the last attitude can now be justified by Bayesian methods with a sufficiently narrow prior, so don’t blame R. A. Fisher for p-value deflation.