FDA, Erlich's escape, EPA, Ignorance, Education
Thanks to the folks at National Institutes of Health for my introduction to statistics and thus experimental design
Working at the Institutes and having cohorts that are serving (is that the correct term) our country as a Doctor by being assigned to the Institutes is not the worst of all worlds. You get to commune with some really great people and maybe you learn a bit as well. The problem begins when work equals research, research involves experimentation and experimentation in turn means experiments; which leads to experimental design, especially if animals are involved.
Now the first question after deciding which species of poor animal will be the subject of your investigations is how many of the critters to order. Money being in short supply, space limited and they smell, you don't want to have more than absolutely necessary. How many is enough. Fortunately having had a thorough course in Statistics (somehow sandwiched in between the course on office management and finance) Medical Doctors are ready to answer this most perplexing question.
If you could use just one animal, it sure would make life easy. You test the chemical (if that is what you are studying(?), on one animal, watch what happens, record the results, excite the press with your new found discovery, rush off to national meetings and join the ranks of the scientifically challenged. But alas, this is not to be. Remembering your statistics, this is a sample population (of rats) you are studying, not the entire universe of rats, so you are forced to use a simplified means of reducing error. You must use the formula, n-1, in your calculations, &c;. Well, when n equals 1, higher math tells us that 1-1=0, so it's no go, you have got to have more animals than one. But how many more? Here I am indebted to that renowned scientist, L. _, who explained the solution in layman's terms to me.
L_ said and I quote, "Never do experiments in pairs, regardless of the rules of statistics. If you make two observations, they, for sure, will not be exactly the same so now you have confusion. Which to believe?" L_'s approach was simplicity in itself. Three results is enough. That's right three, and how did he arrive at this keystone to science? Now let's assume that you have made three numeric observations and they are all over the board. Well scientific intuition tells you that two of the three are in the direction that you want to go, so you simply throw out that wayward one. Can you do this? Of course, statisticians are human after all and they expect to be paid for their sage advice and have found a way around this problem. They call the offending one, an outlier, and off with it. Now your data has a nice fit; go publish your papers.
Oh, by the way, every good bit of science (as described above) has its rewards. Just count the number of papers (publications) you can get from your morning's work. First you rush off to a meeting, and give a stirring talk (In some folk's count, this is paper number one.) Then in one of the quickie journals you can always find an audience and out comes the paper in about a month or so. (Paper number two). With this under your belt it is time for reflection and publishing a longer and more detailed report, one that someone schooled in the art may actually be able to understand and replicate if they are so disposed. (Paper number three). Why with all this excitement, a review is in order (Paper number four). Of course you will be invited to write a chapter in a forthcoming book (Paper number five). Now you have arrived, you can be the author or at least the editor for another book (Don't count this as a paper, it has much more weight and importance, and it sure looks good on the growing list of references.) Let's just say publications number five and we, being modest don't count the endless press articles in the Washington Post, Wall Street Journal, New York Times, Newsweek, &c;.
Now as head of a research lab or perhaps an institute, others are anxiously pursuing your field of endeavor, but there is a dark cloud on the horizon. Some dumb cluck can't seem to get the right answer, if fact is casting dispersions on your work. Unfortunately, he/she is right, so here comes publication number six, a retraction. Don't let this bother you, happens all the time, and since the weight of numbers; five plus, one minus, it is more than likely that the retraction will be missed. And of course in your darkest hour, you have the support of those that followed in your footsteps and published as well.
Sound unlikely? Ask the people at Tulane University.
ABOUT Joe Wortham
Directory