Przeczytaj tekst i wybierz odpowiedź zgodną z jego treścią.
Nothing to Report
I spent the second and third years of my doctorate testing whether a particular chemical compound could slow the growth of certain cells in the laboratory. I designed the experiments with great care, repeated them more times than I can now remember and analysed the results with every method my statistics course had taught me. The answer, in the end, was unambiguous: it could not. The compound did nothing at all. When I took the results to my supervisor, he looked through them, nodded, and said that I had better put them in the drawer. I knew exactly what he meant. Nobody would ever read them.
That drawer, in one form or another, exists in almost every laboratory in the world. Scientific journals have long favoured results that are positive, surprising and clear, and researchers, knowing this, rarely bother to write up experiments in which nothing happened. The consequences are more serious than they might appear. If only the studies that find an effect are published, the scientific record as a whole will tend to exaggerate how often, and how strongly, such effects occur. And because failures go unrecorded, other researchers, quite unaware, may spend months or years repeating experiments that have already been shown not to work.
These problems came to wide public attention when several large projects attempted to repeat well-known studies, first in psychology and then in other fields, and found that a worrying proportion of the original results could not be reproduced. It is tempting to see in this a story of fraud, and there have certainly been a few cases of deliberate deception. But they are the exception. Far more often, the explanation is depressingly ordinary: honest people responding quite sensibly to a system that rewards the wrong things. A researcher whose career depends on publishing striking findings will, without any intention to deceive, tend to pursue and present the results most likely to be noticed.
There are now serious attempts to change this. Increasingly, researchers are encouraged to register their plans, including their hypotheses and the methods they will use to analyse their data, before they begin an experiment. Some journals have gone further, agreeing to publish a study on the strength of its design, before anyone knows what the results will be. These seem to me among the most valuable reforms science has undertaken in decades. Yet their adoption remains patchy, and they sit uneasily alongside hiring committees that still judge young scientists largely by how many papers they have placed in the most prestigious journals.
The deepest obstacle, though, may be emotional rather than institutional. I can still recall the shame I felt over those two years of work, as if the compound's failure to do anything had somehow been my own. Many colleagues have described the same feeling. There is something perverse in this. We train scientists to be endlessly curious about the world and to follow the evidence wherever it leads, and then we teach them, without ever quite saying so, to be embarrassed when the world declines to give them the answer they hoped for.
Several years after I left the laboratory, I finally wrote up my results and submitted them to a journal that exists specifically to publish findings of this kind. The paper was short and, by any normal standard, unremarkable. A few months later, I received an email from a researcher on the other side of the world. He had been about to begin almost exactly the same series of experiments, he wrote, and having read my paper, he had decided not to. I had saved him at least a year. Of everything I have published, it remains, by some distance, the most useful.