Advanced

No evidence of publication bias in climate change science

Harlos, Christian LU ; Edgell, Tim C. and Hollander, Johan LU (2017) In Climatic Change 140(3). p.375-385
Abstract

Non-significant results are less likely to be reported by authors and, when submitted for peer review, are less likely to be published by journal editors. This phenomenon, known collectively as publication bias, is seen in a variety of scientific disciplines and can erode public trust in the scientific method and the validity of scientific theories. Public trust in science is especially important for fields like climate change science, where scientific consensus can influence state policies on a global scale, including strategies for industrial and agricultural management and development. Here, we used meta-analysis to test for biases in the statistical results of climate change articles, including 1154 experimental results from a... (More)

Non-significant results are less likely to be reported by authors and, when submitted for peer review, are less likely to be published by journal editors. This phenomenon, known collectively as publication bias, is seen in a variety of scientific disciplines and can erode public trust in the scientific method and the validity of scientific theories. Public trust in science is especially important for fields like climate change science, where scientific consensus can influence state policies on a global scale, including strategies for industrial and agricultural management and development. Here, we used meta-analysis to test for biases in the statistical results of climate change articles, including 1154 experimental results from a sample of 120 articles. Funnel plots revealed no evidence of publication bias given no pattern of non-significant results being under-reported, even at low sample sizes. However, we discovered three other types of systematic bias relating to writing style, the relative prestige of journals, and the apparent rise in popularity of this field: First, the magnitude of statistical effects was significantly larger in the abstract than the main body of articles. Second, the difference in effect sizes in abstracts versus main body of articles was especially pronounced in journals with high impact factors. Finally, the number of published articles about climate change and the magnitude of effect sizes therein both increased within 2 years of the seminal report by the Intergovernmental Panel on Climate Change 2007.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Contribution to journal
publication status
published
subject
in
Climatic Change
volume
140
issue
3
pages
11 pages
publisher
Springer
external identifiers
  • scopus:85007443342
  • wos:000393744800004
ISSN
0165-0009
DOI
10.1007/s10584-016-1880-1
language
English
LU publication?
yes
id
40847731-5d8c-4077-afe2-bc93a408d5f2
date added to LUP
2017-01-13 07:37:10
date last changed
2018-03-04 04:57:29
@article{40847731-5d8c-4077-afe2-bc93a408d5f2,
  abstract     = {<p>Non-significant results are less likely to be reported by authors and, when submitted for peer review, are less likely to be published by journal editors. This phenomenon, known collectively as publication bias, is seen in a variety of scientific disciplines and can erode public trust in the scientific method and the validity of scientific theories. Public trust in science is especially important for fields like climate change science, where scientific consensus can influence state policies on a global scale, including strategies for industrial and agricultural management and development. Here, we used meta-analysis to test for biases in the statistical results of climate change articles, including 1154 experimental results from a sample of 120 articles. Funnel plots revealed no evidence of publication bias given no pattern of non-significant results being under-reported, even at low sample sizes. However, we discovered three other types of systematic bias relating to writing style, the relative prestige of journals, and the apparent rise in popularity of this field: First, the magnitude of statistical effects was significantly larger in the abstract than the main body of articles. Second, the difference in effect sizes in abstracts versus main body of articles was especially pronounced in journals with high impact factors. Finally, the number of published articles about climate change and the magnitude of effect sizes therein both increased within 2 years of the seminal report by the Intergovernmental Panel on Climate Change 2007.</p>},
  author       = {Harlos, Christian and Edgell, Tim C. and Hollander, Johan},
  issn         = {0165-0009},
  language     = {eng},
  number       = {3},
  pages        = {375--385},
  publisher    = {Springer},
  series       = {Climatic Change},
  title        = {No evidence of publication bias in climate change science},
  url          = {http://dx.doi.org/10.1007/s10584-016-1880-1},
  volume       = {140},
  year         = {2017},
}