Thanks. Of course, I defer to your personal expertise and knowledge. You suggest a mechanism, which is what largely distinguishes science from epidemiology.
I do not, however, withdraw my implied criticism of the original study as reported in the media, which is summed up by the chart I produced last March.
If your hypothesis is correct, an extremal statistic, such as the lowest decile, should have offered a dramatic confirmation. Taking the mean dilutes the effect until it is, to all intents and purposes, invisible. This is because the mean will be dominated by the genetic variants.
I am one of those slightly affected by "Ducks Disease", a term coined by the Welsh comedian Harry Secombe. He would demonstrate it by sitting between two men who appeared to be of the same height. Then they stood up and he was a foot shorter. I am the same height as a six footer sitting down, but four inches shorter standing up. In my case it is purely hereditary, as both parents were shorties.
There is a world of difference between testing a hypothesis by statistical inference and trawling for random correlations. Take the case of fingers. This from Number Watch last August:
To compound the offence The Times used the occasion to fill up a few gaping column inches with a revisit to the finger farce. It is an interesting feature of media coverage of junk science that, with time, the mays, mights and coulds fade away and the results of tacky small surveys become the received wisdom. The differential length of fingers, like the masculinity of faces, is all down to the exposure to testosterone in the womb. How do they know, have they measured it? It now relates among other things to homosexuality, sperm quality, breast cancer, heart disease, autism, dyslexia and left handedness. They left some out (see Fingers back in and also Digital distortion, Fingers out, Return of Finger Man, Return of the finger prince etc.) The permutations of random and accidental correlations are endless.
These random trawls for correlations of body measurements with various conditions, either as part of a large data dredge or as a study so small as to be meaningless, do nothing for science, medicine or the welfare of the human race.