R. E. Benestad1, H. O. Hygen1, R. van Dorland2, J. Cook3, and D. Nuccitelli41The Norwegian Meteorological Institute, Norway 2The Royal Netherlands Meteorological Institute, The Netherlands 3Global Change Institute, the University of Queensland, Australia 4Tetra Tech, Inc., McClellan, California, USA
Received: 04 Apr 2013 – Accepted for review: 22 Apr 2013 – Discussion started: 03 May 2013
Abstract. Replication is an important part of science, and by repeating past analyses, we show that a number of papers in the scientific literature contain severe methodological flaws which can easily be identified through simple tests and demonstrations. In many cases, shortcomings are related to a lack of robustness, leading to results that are not universally valid but rather an artifact of a particular experimental set-up. Some examples presented here have ignored data that do not fit the conclusions, and in several other cases, inappropriate statistical methods have been adopted or conclusions have been based on misconceived physics. These papers may serve as educational case studies for why certain analytical approaches sometimes are unsuitable in providing reliable answers. They also highlight the merit of replication. A lack of common replication has repercussions for the quality of the scientific literature, and may be a reason why some controversial questions remain unanswered even when ignorance could be reduced. Agnotology is the study of such ignorance. A free and open-source software is provided for demonstration purposes.
Benestad, R. E., Hygen, H. O., van Dorland, R., Cook, J., and Nuccitelli, D.: Agnotology: learning from mistakes, Earth Syst. Dynam. Discuss., 4, 451-505, doi:10.5194/esdd-4-451-2013, 2013.