Abstract
We constructed a survey to understand how authors and scientists view the issues around reproducibility, focusing on interactive elements such as interactive figures embedded within online publications, as a solution for enabling the reproducibility of experiments. We report the views of 251 researchers, comprising authors who have published in eLIFE Sciences, and those who work at the Norwich Biosciences Institutes (NBI). The survey also outlines to what extent researchers are occupied with reproducing experiments themselves. Currently, there is an increasing range of tools that attempt to address the production of reproducible research by making code, data, and analyses available to the community for reuse. We wanted to collect information about attitudes around the consumer end of the spectrum, where life scientists interact with research outputs to interpret scientific results. Static plots and figures within articles are a central part of this interpretation, and therefore we asked respondents to consider various features for an interactive figure within a research article that would allow them to better understand and reproduce a published analysis. The majority (91%) of respondents reported that when authors describe their research methodology (methods and analyses) in detail, published research can become more reproducible. The respondents believe that having interactive figures in published papers is a beneficial element to themselves, the papers they read as well as to their readers. Whilst interactive figures are one potential solution for consuming the results of research more effectively to enable reproducibility, we also review the equally pressing technical and cultural demands on researchers that need to be addressed to achieve greater success in reproducibility in the life sciences.
Highlights
Reproducibility is a defining principle of scientific research, and broadly refers to the ability of researchers, other than the original researchers, to achieve the same findings using the same data and analysis (Claerbout and Karrenbach, 1992)
The first survey was conducted in November 2016 and sent out to 750 researchers working in the Norwich Biosciences Institutes (NBI) at a postdoctoral level or above
We reported the qualitative results of the surveys in accordance with the Standards for Reporting Qualitative Research (SRQR) (O’Brien et al, 2014)
Summary
Reproducibility is a defining principle of scientific research, and broadly refers to the ability of researchers, other than the original researchers, to achieve the same findings using the same data and analysis (Claerbout and Karrenbach, 1992). Irreproducible experiments are common across all disciplines of life sciences (Begley and Ellis, 2012) and many other disciplines (Ioannidis, 2005), such as psychology (Open Science Collaboration, 2015), computer science (Crick et al, 2017), economics (Ioannidis et al, 2017; Christensen and Miguel, 2018) and ecology (Fraser et al, 2018). 2012 study showed that 88% of drug-discovery experiments could not be reproduced even by the original authors, in some cases forcing retraction of the original work (Begley and Ellis, 2012). Irreproducible genetic experiments with weak or wrong evidence can have negative implications for our healthcare (Yong, 2015). While irreproducibility is not confined to biology and medical sciences (Ioannidis and Doucouliagos, 2013), irreproducible biomedical experiments pose a strong financial burden on society; an estimated $28 billion was spent on irreproducible biomedical science in 2015 in the United States alone (Freedman et al, 2015)
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call