The Unprofessionalism of Scientists:
There was a recent article (Empty Archives, NATURE vol. 461, no. 7261 September 10, 2009 page 160) with an accompanying introduction called Data’s Shameful Neglect. While they are kicking themselves I thought I’d get with it and join the fun.
The issue is that for a lot of studies the data upon which the conclusion is based is not made readily available to other scientists in a digestible form. And there is the feeling that it should be. At my own level, I quite agree. More than once I have found myself looking at a graph or chart and measuring it, as it were, with my thumbnail to recover the numbers I wanted. It does not make a lot of sense, does it? The graph is there to display numbers. That’s convenient, but it would be nice to have the numbers as well.
In the second Danish paper, there are data points summarized with a line. The graph shows the relationship between the distances between the places where couples were born and their number of children. Of course for my purposes I have an interest in the relationship between fertility and population size. Population size is roughly proportional to area. So I wanted to compare fertility with the square of the distances between birthplaces. That should have been easy. But I wound up having to spend some time with sharpened pencils and a couple old fashioned drafting triangles doing geometric constructions with similar triangles to measure the locations of the points.
I used points on their lines rather than their own points, and their line was basically a fourth order equation. I did not think their data was clean enough nor had enough data points at the apex of the curve to justify changing a third order equation to a forth order, so I moved a single point a short distance, well within the scatter of the original data. That’s honest enough if you admit what you’re doing. But the final result was not totally satisfying. It looked too slick. I routinely find real data that flows so predictable that it is obvious that a single process dominates infertility. But I still did not much like it. I would been far happier to present the original data unmodified except as I explained; it made my point well enough. Some day I hope to show you the result of that effort.
When I was first involved in science as a laboratory assistant in the nearby medical school when I was home from college in the summers, there was a feeling that one should never destroy original data. Calculations could come and go, but once a measurement was made it was kept forever. This did not prevent the technicians on occasion from taking a paintbrush to paint over dust specs on a photographic plate to prepare it for publication. That worried me at the time, but they were only prettying it up, not changing the data.
But in those days you pretty much kept the data in the lab. My older brother, who had done similar work, once remarked that some day they ought to make all lab notebooks available, even ones that had not yielded publishable data. Such an archive could help another scientist avoid going down a blind alley. That was not feasible at the time, but now with digital archiving it not only is possible but there is the strong suggestion that at least the data that has served to produce papers should become public. And this might do more than warn workers away from blind alleys but be the source of new discoveries.
That is a change. Not so long ago there was the sense that “data mining” was improper. It meant that just searching somebody else’s data for a hitherto unsuspected correlation was the lowest form of behavior, as low as moving a data point on a graph (oops!). But there is a technique called “meta-analysis” that has become acceptable. Suppose you have twenty studies that attempt to prove, oh, that aspirin cures headaches. Each study takes a very few people with headaches, gives some aspirin and some a placebo (after proper screening that the pain is not being caused by bleeding into the brain, as aspirin has an effect on clotting time) and compares the outcome. The numbers are small but each study is inconclusive, we shall assume just for argument sake. But if you add them all up, you get a significant effect. That is meta-analysis. Obviously it is strewn with pitfalls, because no two papers are identical, and one must decide which ones to accept into the analysis, posing the danger that the criteria of acceptance might just alter the final outcome.
So data mining, of which as you should know I do a lot of, is not frowned upon so much any longer. This is gratifying.
But they have used the word “shameful,” and there I do have a gripe.
Lawyers are professionals. If you get arrested, you can get a lawyer. No lawyer has to take your case, but generally one will. If you cannot afford one, the state will arrange one for you. Of course any lawyer would make more money working for a large law firm and consulting with giant corporations than trying to help the little guy. But somebody is there for the little guy, too.
Teachers are professionals. If there is something you want to be taught, you can hire somebody to do it. Usually you go to school. But you can also hire a private tutor. There is good money and a lot more prestige in being an exalted professor in a snazzy university and lecturing to a class than in sitting down for hours with the worst, most disorganized, lazy and slow witted member of the same class and trying to help that student pass. But somebody is there for the little guy.
Ministers are professionals. There is prestige in preaching to a large congregation of interested and educated believers. There is less prestige in listening to the spiritual doubts of somebody who has trashed his or her own life by making bad and even ill willed choices. But somebody is there for the little guy.
Doctors are professionals. There is good money and prestige to be gained by doing very difficult procedures to ameliorate dire medical problems. There is less money or prestige in stitching up the face of some drunk who has hurt himself and is not even sure why. I was once working on such a man, who had been brought to the emergency room by a police officer. I listened to the man’s story and then testosterone poisoning cut in and I taunted the cop.
“You didn’t do this to him, did you?”
My intoxicated patient rallied to the truth and said, “No, Doc. He didn’t do it.” (Police are professional; there is a cop for the little guy. And you know something? Sometimes there is even a drunk for the little guy.)
The peace officer smiled serenely and said, “Why no. I did not do it. And I can prove it.”
If I had had sufficient coordination to raise an eyebrow, I would have. But I did look at him questioningly. He fished in his pocket and drew out a handsome black leather object that had been hand stitched. It was a piece of lead sewn between two of leather. It was quite flat so it fit nicely in a pocket.
”This is my blackjack,” he went on in a mock injured tone. “See? It’s flat. It won’t leave a cut.” For the record, the blackjack was pristine. He showed me both sides. He had never struck with it. I hope I acknowledged his verbal parry-riposte with sufficient grace, but this was up North, and the rules are different there.
There is a doctor for the little guy. A doctor does not have to take a case any more than a lawyer does, but if not he is obliged to find a doctor who will. He must make a “proper referral.”
And there are exalted scientists, who command immense resources usually either provided by major corporations, by public tax supported funding or by private endowments. But suppose you are a little guy. Suppose you have a scientific question, something that needs some research done on it. Do you think these mighty of the earth will take an interest, will make a proper referral, will suggest how you might proceed, will even quote you a fee for looking into it? No, they will not. Ten years of experience says they will not. That is unprofessional of them.
There are no scientists for the little guy.
M. Linton Herbert
There have been 2,647 visitors so far.