Dishonest conformity in peer review

If you have found this article useful and you think it is important that researchers across the world have access, please consider donating, to ensure that this valuable collection remains Open Access.

Prometheus is published by Pluto Journals, an Open Access publisher. This means that everyone has free and unlimited access to the full-text of all articles from our international collection of social science journals. Furthermore Pluto Journals authors don’t pay article processing charges (APCs).

Prometheus

Volume 33, Issue 3

Record: found
Abstract: found
Article: found

Is Open Access

Dishonest conformity in peer review

research-article

Author(s): Sotaro Shibayama ^a ^, ^b , Yasunori Baba ^a

Publication date (Print): 1 September 2015

Journal: Prometheus

Publisher: Pluto Journals

Bookmark

Abstract

Honesty in scientific publication is critical for scientific advancement, but dishonesty is commonly and increasingly observed in misconduct and other questionable practices. Focusing on dishonest conformity in peer review, in which authors unwillingly obey referees' instructions in order to have their papers accepted even if the instructions contradict the authors' scientific belief, the current study aims to investigate the determinants of dishonesty. Drawing on survey data of Japanese life scientists, this study shows that the conflict between authors and referees in peer review is common. A majority of scientists follow referees' instructions rather than refute them. The results suggest that conformity occurs more frequently (1) in biology than in medicine and agriculture, (2) when authors are in strong scientific competition, (3) if authors are associate professors rather than full professors, (4) if authors have no foreign research experience, and (5) in low-impact journals rather than in medium-impact journals.

Content

Author and article information

Contributors

Sotaro Shibayama:

ORCID: http://orcid.org/0000-0002-6701-9828

Journal

Journal ID (doi): 10.2307/j50022063

Journal ID (jcode): prometheus

Title: Prometheus

Publisher: Pluto Journals

ISSN: 0810-9028

ISSN: 1470-1030

Publication date (Print): 1 September 2015

Volume: 33

Issue: 3 ( doiID: 10.1080/prometheus.33.issue-3 )

Pages: 215-233

Affiliations

[ ^a ]Research Center for Advanced Science and Technology, University of Tokyo, Tokyo, Japan;

[ ^b ]Department of Technology Management for Innovation, University of Tokyo, Tokyo, Japan

Article

Publisher ID: 08109028.2015.1114745

DOI: 10.1080/08109028.2015.1114745

SO-VID: 0efb088e-9f8c-4f09-8a43-4b3b60130d60

License:

All content is freely available without charge to users or their institutions. Users are allowed to read, download, copy, distribute, print, search, or link to the full texts of the articles in this journal without asking prior permission of the publisher or the author. Articles published in the journal are distributed under a http://creativecommons.org/licenses/by/4.0/.

Custom metadata

lang eng

ScienceOpen disciplines: Computer science,Arts,Social & Behavioral Sciences,Law,History,Economics

Notes

http://www.singaporestatement.org/statement.html. The other three principles are accountability, professional courtesy and fairness, and good stewardship.
http://grants1.nih.gov/grants/research_integrity/whatis.htm. The other shared values are accuracy, efficiency, and objectivity.
http://www.nsf.gov/oig/resmisreg.pdf, http://grants.nih.gov/grants/research_integrity/research_misconduct.htm.
School Basic Survey (http://www.mext.go.jp/b_menu/toukei/chousa01/kihon/1267995.htm).
Fields are determined by WoS subject categories. A complete list of subject categories is available in supplementary data (Table S1 available online at http://dx.doi.org/10.1080/08109028.2015.1114745).
As of 2014, among 86 national universities, 53 have departments related to life sciences (e.g. medicine, science, agriculture).
We set this condition to spare the effort of preparing questionnaires in foreign languages. Comparing foreign-borns and Japanese scientists is of interest, but because the proportion of foreign-born scientists in Japan is extremely small (Franzoni et al., 2012), meaningful comparison would have been difficult.
We first grouped the authors into three scientific fields (medicine, agriculture, and biology) based on WoS subject categories. Then we employed stratified random sampling in each field. For the JIFs, since we anticipated that authors of high-JIF papers might be relatively unwilling to respond because of being busy, etc., we oversampled high-JIF authors. As we were also interested in the behavior of low-JIF authors, they, too, were oversampled.Sampling weights of 5.0, 1.0, and 2.5 are given to top, middle, and bottom JIFs, respectively. For university ranks, sampling weights of 1.0, 1.5, and 2.0 are given to Tiers 1, 2, and 3, respectively. We oversampled lower-ranked universities because their population is smaller.
Japanese academia is highly gender-biased. Geuna and Shibayama (2015) find that the proportion of female professors in STEM fields is between 2% and 8%.
If they had multiple instances, we asked them to choose the specific case which revealed the most serious inconsistency.
Unwillingly adding experiments/data may not violate the honesty norm, although we still believe that adding unnecessary experiments/data is considered questionable in that it is a waste of resources.
We used the Journal Citation Reports of 2012, published by Thomson Reuters, owners of Web of Science.
The interpretation requires caution because of small sample size and selection bias.
When two papers report similar results, the first published paper is given greater reputational reward than the second published paper (Merton, 1973). We suppose that the difference of reward given to the two papers is larger in basic fields than in applied fields.
The field of medicine includes both clinical and non-clinical research. We tested whether propensity to conformity differs between them, but did not find a significant difference.
However, when we examined the effect of the size of network (e.g. the number of coauthors, etc.), we did not find a significant effect.
For example, interaction among multiple authors is relevant to authorship abuse; interaction between the principal investigator and students may be relevant to data fabrication.

Comments

Comment on this article

[1] http://www.singaporestatement.org/statement.html. The other three principles are accountability, professional courtesy and fairness, and good stewardship.

[2] http://grants1.nih.gov/grants/research_integrity/whatis.htm. The other shared values are accuracy, efficiency, and objectivity.

[3] http://www.nsf.gov/oig/resmisreg.pdf, http://grants.nih.gov/grants/research_integrity/research_misconduct.htm.

[4] School Basic Survey (http://www.mext.go.jp/b_menu/toukei/chousa01/kihon/1267995.htm).

[5] Fields are determined by WoS subject categories. A complete list of subject categories is available in supplementary data (Table S1 available online at http://dx.doi.org/10.1080/08109028.2015.1114745).

[6] As of 2014, among 86 national universities, 53 have departments related to life sciences (e.g. medicine, science, agriculture).

[7] We set this condition to spare the effort of preparing questionnaires in foreign languages. Comparing foreign-borns and Japanese scientists is of interest, but because the proportion of foreign-born scientists in Japan is extremely small (Franzoni et al., 2012), meaningful comparison would have been difficult.

[8] We first grouped the authors into three scientific fields (medicine, agriculture, and biology) based on WoS subject categories. Then we employed stratified random sampling in each field. For the JIFs, since we anticipated that authors of high-JIF papers might be relatively unwilling to respond because of being busy, etc., we oversampled high-JIF authors. As we were also interested in the behavior of low-JIF authors, they, too, were oversampled.Sampling weights of 5.0, 1.0, and 2.5 are given to top, middle, and bottom JIFs, respectively. For university ranks, sampling weights of 1.0, 1.5, and 2.0 are given to Tiers 1, 2, and 3, respectively. We oversampled lower-ranked universities because their population is smaller.

[9] Japanese academia is highly gender-biased. Geuna and Shibayama (2015) find that the proportion of female professors in STEM fields is between 2% and 8%.

[10] If they had multiple instances, we asked them to choose the specific case which revealed the most serious inconsistency.

[11] Unwillingly adding experiments/data may not violate the honesty norm, although we still believe that adding unnecessary experiments/data is considered questionable in that it is a waste of resources.

[12] We used the Journal Citation Reports of 2012, published by Thomson Reuters, owners of Web of Science.

[13] The interpretation requires caution because of small sample size and selection bias.

[14] When two papers report similar results, the first published paper is given greater reputational reward than the second published paper (Merton, 1973). We suppose that the difference of reward given to the two papers is larger in basic fields than in applied fields.

[15] The field of medicine includes both clinical and non-clinical research. We tested whether propensity to conformity differs between them, but did not find a significant difference.

[16] However, when we examined the effect of the size of network (e.g. the number of coauthors, etc.), we did not find a significant effect.

[17] For example, interaction among multiple authors is relevant to authorship abuse; interaction between the principal investigator and students may be relevant to data fabrication.