Articles to 2017-09-08

Zum Seitenende      Übersicht Artikel      Home & Impressum

First the link to this week’s complete list as HTML and as PDF.


Reading Cobey and Moher et al. it seems to me, that predatory journals and their machinations are among the least of our problems. According to the authors, nearly all of the articles they have looked at were seriously flawed and quite a few came from some of the most prestigious universities, written or at least supervised by senior scientists employed there. Were it not for those shoddy publishers, those articles might well have ended up in the reputable journals. In fact the authors clearly find just that. The fraction of badly flawed articles is much lower there, but very far from zero.

A tiny bit off the tip of that iceberg is regularly flagged here and those are only the ones, whose faults are obvious at first sight even to me. When not even places like Harvard care about their academic standards, what can we expect from the rest of the publishing avalanche? Take a close look at what is still cited and valued decades later, at how many articles those authors ever wrote, and how long the ones still read took to germinate, and the difference becomes very clear. I agree with the authors that something, if unchecked, “will continue to erode the integrity of scientific scholarship”, but it’s not predatory journals that first come to mind.


Drummond & Fischhoff are another example where we are shown no data and no errors but regression lines only. It seems, it can’t be repeated often enough, that any regression run on any random data at all will always yield a definite result with a non-zero coefficient and about one in twenty of those results will turn out statistically significant – usually many more than that, as the statistical test is often sloppily applied.

Yes, most of the results in their fig.1 can be given a plausible explanation. Being plausible does not make something true. It took a very long time for the scientific method finally to supersede all those plausible and convincing scholastic explanations of the world and we are just now experiencing the end of that scientific and critical era.

Let’s just look at climate change vs. scientific literacy, according to table 1 among the most highly significant of all results. The regressions all cross near the middle of the range. So scientifically illiterate people tend to espouse the exact opposite views to those of their political peer group? Really?

Feeding a sackful of meaningless data into some statistics package or other and randomly pressing knobs until some result seems to look nice is not science. It may be a first step towards generating a hypothesis and devising a test for it, but no more.


Voosen is just one more example where carbon dioxide concentrations from ice cores are at odds with other proxies. Of all gases CO2 is about the most soluble in water and the most prone to diffuse out of a compressed bubble. Admittedly the existence of C4 plants and their relative dominance yields strong support for low values in the cold periods. The question is, just how low?


Ethier et al. is another example for the early importance of dairying and one more refutation of the secondary products revolution.

Zum Anfang      Übersicht Artikel      Home & Impressum

Creative Commons Attribution-Share Alike 3.0 Unported License Viewable With Any Browser Valid HTML 4.01! Valid CSS!