Wansink Apologism

[bob sterman]

Oh dear....

Massaging data to fit a theory is not the worst research sin
https://www.timeshighereducation.com/op ... search-sin

[Martin Y]

The subtitle torpedoes the title claim: "Massaging data to fit a theory is not the worst research sin - The furore over Brian Wansink’s data handling overlooks the fact that science should above all be about explanation, argues Martin Cohen"

It's one thing to talk about science being about explanation, although that's highly debatable, but it's absolute b.llsh.t to say research is about explanation. "Massaging" research data is precisely the antithesis of science. It's making up stories you like. It's the very reason science had to be invented, to stop us kidding ourselves.

[Stupidosaurus]

Not seeking to agree or justify, but hasn't there been some evidence that some of science's historical leading lights have beautified their data in the past? History written by the victors blah blah. I would dig out an example but I'm lazy and about to watch some sh.tty TV.

[basementer]

Not seeking to agree or justify, but hasn't there been some evidence that some of science's historical leading lights have beautified their data in the past? History written by the victors blah blah. I would dig out an example but I'm lazy and about to watch some sh.tty TV.

ISTR Mendel's results from plant breeding were good, good to the point of implausibility.

[shpalman]

Not seeking to agree or justify, but hasn't there been some evidence that some of science's historical leading lights have beautified their data in the past? History written by the victors blah blah. I would dig out an example but I'm lazy and about to watch some sh.tty TV.

https://en.wikipedia.org/wiki/Oil_drop_experiment

[Boustrophedon]

So if it turns out to be true then it's beautification but if it turns out not to be true it's fraud? How far from the truth was Cyril Burt?

[Martin_B]

Not seeking to agree or justify, but hasn't there been some evidence that some of science's historical leading lights have beautified their data in the past? History written by the victors blah blah. I would dig out an example but I'm lazy and about to watch some sh.tty TV.

ISTR Mendel's results from plant breeding were good, good to the point of implausibility.

Wasn't it that the people counting the results (not Mendel himself) knew the results which Mendel expected, and fudged the results.

[Bird on a Fire]

Not seeking to agree or justify, but hasn't there been some evidence that some of science's historical leading lights have beautified their data in the past? History written by the victors blah blah. I would dig out an example but I'm lazy and about to watch some sh.tty TV.

ISTR Mendel's results from plant breeding were good, good to the point of implausibility.

Wasn't it that the people counting the results (not Mendel himself) knew the results which Mendel expected, and fudged the results.

Yes, apparently his monk underlings are the suspect culprits.

[bmforre]

Mendel's underlings certainly knew they must not spread falsehoods against their neighbours.
But probably didn't regard banana flies as neighbours whether those lived in a cloister garden or not?

[Stupidosaurus]

Yes, Mendel's peas and the oil drop experiment. That's exactly what I had in mind. Definitely. Although fruit flies were done by Morgan the Organ, I think. He played the hymns at Mendel's monastery.

[bob sterman]

First I must apologise for not noticing this "Wansink Apologist" article is actually from more than a year ago. For some reason it popped up on the THES page when I was browsing.

And there is already a good reubuttal here...

https://neuroneurotic.net/2018/12/08/ma ... o-science/

But this was probably discussed at length back on the old forum.

[Allo V Psycho]

So if it turns out to be true then it's beautification but if it turns out not to be true it's fraud? How far from the truth was Cyril Burt?

Warning: I once wrote an article on Cyril Burt.

How far from the truth was Cyril Burt? tl:dr As far as you can get in the opposite direction. He invented data that contributed to harm, both to his field and to social structures in the UK.

OK. First of all, I don't think science is about 'truth'. I'm not sure humans could ever even understand the 'truth' about the Universe. What I think science does is to produce 'conditional statements of probability that are useful'. (Useful doesn't just mean it generates electricity, or profits). The currently accepted paradigm in any field will have changed in the past, and may well change in the future. The process by which the current paradigm is arrived at, is essential: it goes to the usefulness of the current version, and the likelihood that it can be replaced by something even more useful.

By making up his separated twin studies, and lying about it in his papers, Burt constructed an apparently convincing narrative that still haunts our society today, and incidentally contributes to social inequality. And he always intended to. In his very first paper, he tested the idea that intelligence is inherited, by comparing the intelligence of children to that of their parents, and observing a strong correlation. Except he didn't. He measure the intelligence of a rather small number of children, and correlated it, not with the measured intelligence of their parents, but with the intelligence he inferred from their social status.. In other words, an honest account of the experiment could just as well have come to the opposite conclusion: that intelligence as measured by tests in children is a function of their family's social position. He devoted almost all his career to continuing to fantasise about the strong inheritance of intelligence, and to make up data to convince others. (Oh yeah, he also believed in telepathy, and spirit voices from beyond the grave).

Don, I don't know if you were thinking, 'he wasn't far from the truth, because intelligence is largely inherited'. But what if that paradigm changes? What if correlations between twins and children and parents are due, not to the genetics of inheritance, but to the influence of the uterine environment? In other words, epigenetic, rather than genetic? That would be useful to know. But it seems absurd to think of Burt the Fraud having been somehow 'close to the truth' and then becoming more distant from the truth again as our understanding improved.

In moving beyond the current paradigm, understanding the ways in which the current paradigm has been constructed are essential. You need to know where the wonderfully convincing world of Newtonian mechanics, and classical physics, which had been 'proved' to be 'true' by a million everyday observations, could be overthrown. In that careful review of the past evidence, looking for the crack from which quantum physics is going to emerge, faked data is a disaster, because it may make it necessary to repeat observations, and you don't even know which experiments to repeat.

The more binary we make an outcome, the more likely any charlatan is to 'be correct'. Is Trump 'wise', if some of his policies have un-predicted benefits? No. The process by which a decision is made is a key part of the value of the decision. Fake data always corrupts the process.

Sorry. But I did warn you...


When you make outcomes binary, then the lies of a dishonest scientist might randomly fall within the current paradigm.

[Bird on a Fire]

Very interesting post, thanks AvP.

I'd like to ask about this bit:

What if correlations between twins and children and parents are due, not to the genetics of inheritance, but to the influence of the uterine environment? In other words, epigenetic, rather than genetic? That would be useful to know.

What would be the impact of this shift in paradigm? Obviously in most cases the uterine environment a kid was in was their biological mother's, so what would be changing is the mechanism of heritability. Is the current thinking still that epigenetic mechanisms often reflect not just the mother's inherent state but also environmental factors, stress etc. - in which case we'd be looking at a mixture of biological and social effects masquerading as purely biological? Or am I way off (neither developmental biology nor humans are my area ).

[Allo V Psycho]

Very interesting post, thanks AvP.

I'd like to ask about this bit:

What if correlations between twins and children and parents are due, not to the genetics of inheritance, but to the influence of the uterine environment? In other words, epigenetic, rather than genetic? That would be useful to know.

What would be the impact of this shift in paradigm? Obviously in most cases the uterine environment a kid was in was their biological mother's, so what would be changing is the mechanism of heritability. Is the current thinking still that epigenetic mechanisms often reflect not just the mother's inherent state but also environmental factors, stress etc. - in which case we'd be looking at a mixture of biological and social effects masquerading as purely biological? Or am I way off (neither developmental biology nor humans are my area ).

Yes, that about it. Wikipedia on the Barker hypothesis isn't bad (and I once was a developmental biologist, so that's where I got into it). I'll copy a bit from the Wiki, because it might possibly have some current relevance...

Pregnancy outcomes can impact the wellbeing of a society. Comparisons between the children who were in gestation during the 1918 flu pandemic and those in gestation immediately before or after the health crisis show marked differences between the two groups on census data. Across all socioeconomic measures, those who were fetuses during the crisis attained lower educational achievement, income, and socioeconomic status. Specifically, individuals affected were 15% less likely to graduate high school, 15% more likely to be poor, and 20% more likely to be disabled as adults. Even federal welfare payments were higher for the gestational cohort than those born before or after the flu hit.[2] The same economic researcher, Douglas Almond, has investigated other historical situations affecting particular cohorts of fetuses: children born during or immediately following the Chernobyl nuclear disaster explosion, and China's Great Leap Forward (which resulted in a deadly famine). Both prenatally exposed groups suffered lower cognitive abilities and reduced employment levels.[2] Such outcomes can have lasting impacts on the productivity and economic security of a society for an entire generation of individuals, and perhaps even continue to affect future descendants through changes in gene expression.

DNA dominance is the current paradigm (along with the political paradigm of 'The poor deserve it"). When I was at Primary School, they made us sing:

"The rich man in his castle, the poor man at his gate,
God made them high and lowly, and ordered their estate".

Now the privileged just substitute 'DNA' for God, and here we are.