Reproducibility crisis – science, we have to talk

Under the provocative title “Why Most Published Research Findings Are False”, John Ioannidis, professor at the University of Stanford, analyzed why most of the releases published were wrong. Since then, the health scientist and statistician has made controversial statements about Covid-19 pandemic. However, his analysis of the problems of science business is still highly topical – as two recently published studies from very different specialist areas show.

For almost a year I repeated the same experiment day after day – without success.

–

The concept of “replication crisis”, in German reproducibility crisis, emerged in the early ten years in the inner science discussion in the course of the observation that already published, some of which did not meet the basic criterion of empirical science: It was not possible to reproduce its results in independent experiments. As a basis for applications such as medication development, for example, not reproducible study results are worthless.

The author of a recently published Study wanted to check whether behavioral ecological insect experiments are also affected. Researchers at the Universities of Jena, Bielefeld and Münster had carried out the same three tests in three different laboratories. They tested the effects of lack of feed on defense reactions in leaf wasp larvae, the connection between the body color and the preferred ambient color for grasshoppers and the choice of habitat in rice flour beetles. It was shown that reproducibility was greater than in other research fields, but here too, not all results between the different laboratories agreed. The author suspects the causes that, among other things, differences in the biological heterogeneity of the test animals examined played a role.

The results of the grasshopper experiments were similar between the laboratories of the current study, but could not confirm the original study of 2021. Researchers had shown here that green or brown grasshoppers prefer to stay in environments according to their coloring. In the new experiments, this effect could not be detected. Outside of a special reproducibility study, however, it would hardly be possible to publish this knowledge. “Negative results” and repetitions of already published tests are uninteresting for scientific specialist publishers and are usually not supported with research funding.

Only shortly after Ioannidis published his article – which unfortunately only met me on the Internet years later – I worked in a renowned neuroscientism laboratory. Here I should repeat an already published behavioral attempt with honeybees in order to gain new knowledge. For almost a year, I repeated the same experiment day after day as a student – without success. The results were simply not reproducible. Finally, my data ended up in the trash can and the rest of the science community was not informed that the results of the original study could not be reproduced and the hypotheses based on them could possibly be wrong.

In the next career step, I researched cancer genetics. The research group had just received a funding for a new topic that was the time. For almost two years I tried to reproduce results from various studies as the basis for my doctoral project. In the meantime, the pressure on me was to find ahead quickly and to satisfy my doctoral supervisor, after all, she would graduate and decide on contract extensions. Such research environments cause people to over -interpret their own results, not to take it so precisely with the statistics or even to fake them. However, I just gave up at some point – unfortunately it was out of the question to publish my failures for my supervisor.

In a survey by the renowned specialist journal »Nature«, more than 70 percent of the scientists surveyed in 2016 replied that they had tried unsuccessfully to confirm the results of other researchers. Over half believed in the existence of a reproducibility crisis. Various initiatives worldwide formed to carry out large reproducibility studies in various scientific disciplines. They undermined the suspected difficulty of repeating published results, with numbers in such different science fields as medicine, psychology, economics and social sciences. In April this year published The Brazilian Reproducibility Initiative The results of the six -year attempt to reproduce the processes from 60 biomedical articles. With a devastating outcome: Depending on the rigor of the criteria, the initiative of over 50 research teams only managed to confirm between 15 and 45 percent of the originally published results.

The unsuccessful attempts to reproduce results waste the lifetime of scientists, eat financial, sometimes tax -financed resources and cause unnecessary animal suffering. However, more is at stake in medical research. Incorrect research approaches that are kept alive by false incentives in the science system can endanger human life in clinical studies and lack therapy development for decades.

Today’s time in which science is politically attacked and long -proven findings such as man -made climate change are doubted does not seem to be the right moment for open debates. In the end, there will be more damage to doubts and self -criticism. We therefore urgently need to talk about scientific uncertainties, precarious working conditions and false incentives.

Dr. Isabelle Bartram is a molecular biologist and employee at Gen-Ethical Network e. V.