Articles to 2020-09-09

Zum Seitenende      Übersicht Artikel      Home & Impressum

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

***

Hamer et al. do it again: no data, no spread, no variability, no hint of predictive value, just regression coefficients. And with one single exception those are given as dimensionless numbers with no hint of the – different and incomparable – abscissal units used. So in model 2 the added risk is 27 % per standard deviation in obesity. It is 60 % for maleness which in reality, unless you define female as normal, means 30 %, or rather ± 30 %. For smoking, seemingly treated as yes/no, it's 63 % and for inactivity it's reported as 52 %. 52 % per what? Is is yes and no? If so, what counts as inactivity? Apparently they have registered at least five levels of activity. They also registered at least five categories for alcohol consumption but only report a 61 % higher risk for abstainers. So drinking is good, regardless of the amount? Am I safe, or what?

So what do we see contrary their headline and abstract? Being obese is one of the smaller risks registered. Being white, non-smoking and beer loving is good for me, though I ought perhaps to get out of my easy chair every now and then. Losing weight wouldn't gain me much but a transgender registration would do a lot. Is the study really as meaningless as all that? No it isn't, but the reporting is.

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!