Column: ”Cheating with pea flowers and does it matter whether you are right?”
The monk Gregor Mendel cracked the code for how different genes are inherited in two separate chromosomes. He proved that if you cross a flower with two dominant traits (pink flowers) and a flower with two recessive traits (white flowers), all flowers in the first generation will be pink because they all carry a dominant pink trait. The white flowers appear again in the next generation, but in a smaller proportion, because the traits are recessive.
The experiments with the pea flowers, which changed colour over several generations, became the first important piece of the puzzle in modern genetics. Unfortunately, Mendel did not live to experience his own fame but died in 1884, unaware that he would forever be referred to as the father of genetics.
In the experiments that Mendel carried out, as in all research in which you study outcomes in a population, you expect variability, quite naturally. Even if, on a theoretical basis, one expects a certain proportion of white and pink flowers, one very rarely observes a number that exactly corresponds to the number called the expected value in statistics, the result you would see if there was no noise.
For almost 100 years, statisticians have wondered whether Mendel cheated with his research. The fact is that when you look at Mendel’s collected research results, there is far too little disturbing noise, and the numbers are far too evidently supporting his hypothesis.
Whether the hypothesis is correct or not, some variability should occur. Was he so convinced that he was right, which he was, that he allowed himself to fudge the numbers to prove his hypothesis more convincingly? One reason why we might forgive Mendel is the fact that he was right. With all the facts to hand, that is how many of us feel; it was so obvious that he was right! And from an ethical perspective, his research was harmless, as no human lives were at stake.
However, play with the idea; what if Macchiarini had been right? What if it had been possible to form new epithelial layers on a site skeleton and his operations had been successful? Perhaps the cheating with missing experimental animal data had never been discovered, and the colleagues at Karolinska Institutet who blew the whistle would have had an even harder time being heard. If anyone had ever listened, considering how few did listen even though they were right. Would an eventual success have made the cheating more acceptable?
Of course not, we can all agree on that. By definition, research means that we cannot or should not speculate about the outcome in advance. And if you actually are lucky and land a successful outcome, this doesn’t justify any possible shortcuts.
We work in the world’s most important industry in which life and death are always at stake. This means that we need to be brave and innovative in our research, but we must never, ever take shortcuts that put lives at risk and compromise research ethics. However, I would actually like to forgive Mendel in particular, if nothing else, because he created entertainment for several generations of statisticians.
Footnote:
This column is written in memory of statistics professor Åke Svensson, who died shortly after New Year’s. He peppered many of my years as a statistician with anecdotes like this one about Mendel. Rest in peace, Åke, and thank you for all our inspiring conversations!
Artikeln är en del av vårt tema om News in English.
