Irreproducibility of scientific results has many sources. The three main causes are deliberate falsification of data, publication of incorrect data, or valid results that are simply difficult to replicate. In their editorial in Angewandte Chemie, Robert G. Bergman, University of California, Berkeley, USA (pictured left), and Rick L. Danheiser, Massachusetts Institute of Technology (MIT), Cambridge, USA (pictured right), discuss the topic of irreproducibility in chemical research. The researchers give some recommendations for minimizing this problem, in particular in the area of organic chemistry.

According to the editorial, the deliberate falsification of data is rare, but modification of data such as removing peaks in NMR spectra or only reporting the best yield for a reaction is more common. Investigator bias, where researchers consciously stick to their interpretations or theories or unconsciously interpret their results according to their expectations, is also possible. Furthermore, some research is difficult to reproduce. For example, one in thirteen procedures submitted to Organic Syntheses cannot be reproduced in another laboratory.

How can irreproducibility be minimized? First of all, principal investigators have the responsibility to stay in regular contact with co-workers, to examine data, to ensure that results can be reproduced, and to address issues of reproducibility in a professional manner. Journals should have a requirement for adequate experimental detail in the supporting information, and reviewers should check whether these details are accurate or not.

• Reproducibility in Chemical Research,
  Robert G. Bergman and Rick L. Danheiser,
  Angew. Chem. Int. Ed. 2016.
  DOI: 10.1002/anie.201606591