Improving the Quality of Toxicological Research Findings Using Modern Principles of Reproducible Research

Reproducibility represents the foundation of scientific work and publications, and the materials and methods section in each published article should allow any researcher to repeat the experiment in question and get the same or similar results. Nevertheless, in most scientific papers the data analysis procedure is rarely described well, and it often contains just the basic information on statistical procedures performed. We present all of the basic steps in doing reproducible data analysis, with all the advantages and disadvantages over the non-reproducible methods, on a case study of pesticide exposure and risk assessment. Data is imported from multiple sources (text, excel, access database), and basic description of acquired data, visual and numerical comparison between groups, and modelling of data acquired in real-life studies of pesticide exposure in agriculture are presented. The final products of the data analysis process, tables and figures which are ready for the revision process, are compiled using the R Language and Environment for Statistical Computing and additional packages. Considering the more strict requirements for funding and the increased competition, as well as the slow (but certain) move towards open access, open review and data exchange, doing data analysis the reproducible way will become inevitable in toxicology, as well as other scientific fields. Popularization and training on using free statistical and reproducible research tools should be a priority for young researchers entering this field, as this will result in the improvement of the quality of toxicological research, leading to easier publishing.