Tuesday, September 14, 2010

Axen, Axen (3) and Hawks

John Hawks put up this excellent quote in his post James Randi on scientists

From Randi, J. 1988. "The detection of fraud and fakery." Cell Mol Life Sci 44:287-288:

"Scientists are very easily deceived. They think logically, extrapolate possibilities from evidence presented, assume (with a good probability of being right) certain aspects of the observed data and draw upon their past experience in coming to decisions. This is to say that they act very much as all humans do, struggling with sensory input to derive new facts from it. But scientists do this with a certain authority and certainty born of their training and discipline. They are thus excellent candidates for being flimflammed by a clever operator who is aware of the fact that scientists seldom bring the human element into account."

Axen and Axen do not do either fraud or fakery. Their data are real. But the human element is essential.

Peter

11 comments:

Emily Deans, M.D. said...

The human element drives most scientists crazy except for those in my profession (psychiatrist) or similar - which is, I believe, why hard scientists consider us to be not quite scientists.

Ned Kock said...

Hi Peter.

I have a question for you. A little off-topic. I have been analyzing some data from the China Study II:

http://healthcorrelator.blogspot.com/2010/09/china-study-ii-cholesterol-seems-to.html

http://healthcorrelator.blogspot.com/2010/09/china-study-ii-wheat-flour-rice-and.html

There are two columns (among many) in that data, one for HDL cholesterol, and the other for Apolipoprotein A. They have different values, with the Apo A column being generally higher. Both are provided in mg/dl.

My question is this. If I want to analyze the relationship between HDL (generally speaking) and other variables (e.g., cardiovascular disease deaths), which one (HDL-C or Apo A) would be the most appropriate?

Peter said...

Hi Ned, I think it might be better to ask DrBG on this one, she has a much broader experience of lipoprotein subtypes than I do, especially from her experience at TYP. The simple trite answer is another question: Which sort goes up with animal fat intake? That's the good sort! I have memories that large HDLs are bad-good cholesterol and small are good-good cholesterol particles but I have trouble keeping up with the goods and the bads here. So it might be a ratio of the two, the total number of ApoA molecules, high is good, but better if they are small, high total cholesterol with an ApoA label is bad, especially if there aren't many good particles and they have lots of bad content. I think I'm wandering here. Better go cook some lamb fat.

BTW I'd just comment on TC being low in China: How about sucrose intake being low? And another BTW, did Campbell look at any of the risk factors for heart disease? I'm thinking about HbA1c in particular. I guess not.... Sorry not a lot of answers but I did get the garage largely packed and a mortgage application in through a 2 hour phone call today, productive day.

Great posts though, keep them coming!

Emily, hmmmm, possibly, but the term scientist does tend to get bandied around when people mean clinicians. They are quite different animals, with quite different (and useful) functions. A combination of the two is a rare breed indeed, but they clearly exist!

Peter

Ned Kock said...

Thanks Peter, I will ask G.

Good points.

I am trying to use the measures that I feel I can trust - e.g., two or more measures that seem to load on the right factor, in a factor analysis.

There is a measure of glucose, but it looks very unreliable. It seems that they measured glucose levels at different times for different people, and averaged the results. So we have the average blood glucose for a county.

And you get things like 40 mg/dl! That must be due to error. For that to be an average blood glucose for a county, my guess is that you would have to include quite a few dead people in the sample.

Pål Jåbekk said...

The mad professor
Had a hole in his head
A big black ball of nothing
Which left him there, dead.

The cognitive dissonance
Ate away his brain
The caged rats watched him
Slowly go insane

With the professor dead
The rats broke free
But alas! A caveat
The rats did not see

Quickly the rats scurried
Across the lab room floor
Running for freedom
Running for the door

But the rats all died
The run their demise
Their Sprague-Dawley hearts
Could not support their size

For you see
A professor’s rat
Is many things
But mostly, it is fat

The dead rats
Were fed from a bowl
Hidden in the dark corner
A big black bowl of Crisco oil

Emily Deans, M.D. said...

(I should clarify I meant the academic psychiatrists, which I am only peripherally).

Stan (Heretic) said...

Hi Peter,

I agree about the human element but I disagree with the first statement about your A&A "scientists".

If you see people acting funny, intelligent educated and abusing logic then chances are it is about money. If it is about money then the "F" word may be warranted, however improbable, even though it violates our personal ideas about integrity. If it is not about money then some more sinister far reaching conspiracy, which I doubt.

Another possibility is the human behavioral regression: some humans revert from nomadic-individualistic mindset to herd-like mentality where typical human traits such as creativity, adaptability, curiosity and logic may no longer seem important (to the regressors). They still are intelligent as the rest of us, except they do not use it the way we do, rather it becames a tool of social interaction exclusively.

Digression.

I found uncanny accurate depiction of the type of behavior you are writing about, in Ayn Rand books. Especially Atlas Shrugged. For me, it felt like my own life todate documentary rather than social fiction. It's a part of the "human element" education - a must-read!

Stan (Heretic)

blogblog said...

The reality is that not all science is equal. In my own two decade experience as a scientist the intellectual rigour of various disciples varies immensely. Analytical chemistry, for example, is around 9.5/10 for intellectual rigour compared with perhaps 4/10 for nutrition or exercise science.

The indisputable fact is that physicians are not properly trained to design and supervise complex research projects. Yet the vast bulk of health research is directly controlled by the medical profession via supervision or funding.

Peter said...

Hi Stan,

I'd heard Atlas Shrugged was a good read, it's on my list now.

I would be loathe to use the F word but I certainly would not describe A&A as scientists. Clever operators, yes.

It's then up to anyone to decide whether the 2010 paper is a flimflam even though no one made up any data (as far as I can tell).......

blogblog, my wife is a scientist. She finds working with clinicians somewhat frustrating!

Emily, do you do people watching? It's a very interesting science but perhaps hard to quantify...

Ned, "you would have to include quite a few dead people in the sample." made my day. I keep giggling, somewhat hysterically!

Great stuff Pål, more grinning.

Peter

Emily Deans, M.D. said...

I think the appropriate question is the one often asked on The Daily Show with Jon Stewart - are they stupid, or evil?

Peter said...

Emily, absolutely! The stupid AND evil ones are easy to spot.

Peter