Publish or Perish: Philosophy and Practice
Dennis M. Bier, M.D.
Editor-in-Chief
The American Journal of Clinical Nutrition
Disclosures • In my youth, I was guilty of essentially every indiscretion
that I will discuss.
• I have consulted for, and received payment from, several
governments, various NGOs, academic societies,
disease-oriented associations & wide variety of food and
drug companies. I am completely conflicted.
• I won’t mention any Industry products. Academic
products are more than sufficient to make my case.
Don’t worry. Your paper will be published.
Any study can get published.
> 8,700 journals; ~5,000 with Impact Factor >1
The issue is not getting published, but getting
published where you want your work to appear.
Little evidence-based data on how to achieve this
Lots of eminence-based data, like mine.
>15,000 manuscripts as editor of The AJCN.
Remember, expert opinion is at the bottom of the
evidence-based hierarchy.
Bier’s Publication Principles
A publication is only as good as the
study it reports.
The quality of a study is proportional to
the quality of the question.
“Often, the approximate answer to the right question is preferable to the
precise answer to the wrong question.” (John Tukey)
Bier’s Publication Principles
Frame the question as an hypothesis.
This will simplify writing the paper.
Hypothesis = simple declarative sentence.
“Yes” or “No” answer
“No” answer is as important as “yes” answer
All the easy questions have been answered
Hypothesis-Driven Manuscript Architecture
Introduction
States hypothesis; explains how you derived it.
Methods
Details how you tested your hypothesis
Results
Provides the raw, un-interpreted data
Discussion
Explains whether the data answer the question
Discusses limitations and implications of findings
How To Write Science [Fiction] Stories
There is only one editorial dictum – scientific
accuracy.
[The story] must be an exposition of a scientific
theme and it must be also a story. As a story, it
must be interesting.
Don’t fall into the misapprehension that, because
your story has plenty of science in it, a plot is
therefore unnecessary.
Hugo Gernsback Writers Digest 10:27-29, 1930
Selecting The Journal
Define audience you want to reach.
Choose the best journal in the field.
Your work should always be highest quality.
You generally get the most rigorous review
Reach most discerning & influential readers
Paper will have the highest visibility
Best support for grants and promotion
Most satisfying, personally & professionally
Journal Impact Factor
Measures frequency that an average article is cited.
A running two-year average
All 2015 citations of articles published in 2013 & 2014
divided by sum of all papers published in 2013 + 2014
Reflects currently perceived scientific “value” of a
journal’s articles and, indirectly, of the journal itself.
It does not measure the “value” of your paper.
Selected 2015 Impact Factors NEJM (59.6)
Nature (38.1)
Lancet (44.0)
Science (34.7)
JAMA (37.7)
BMJ (19.7)
Lancet Diabetes/Endo (16.3)
J Clin Invest (12.6)
Diabetes Care (8.9)
Diabetes (8.8)
J Clin Endocrinol Metab (6.2)
Annu Rev Nutr (6.8)
Am J Clin Nutr (6.7)
Nutrition Rev (5.6)
Adv Nutr (5.4)
Internat J Obesity (5.3)
Proc. Nutr. Soc (4.7)
J Nutr (3.9)
Obesity (3.6)
Eur J Nutr (3.5)
J Parenteral Enteral Nutr (3.5)
Brit J Nutr (3.3)
2015 Journal Citation Reports
Relative Ranks of 8,757 Journals Impact Factor
15.0
9.0
6.0
5.0
4.0
2.0
1.0
Percentile
1.0
2.5
5.0
7.0
11.0
36.7
66.6
How Are You Doing?
Science Citation Index
Other indices H-Index, G-Index, Age-Weighted Citation Rate, etc.
Publish or Perish Software and Book
www.harzing.com/resources/publish-or perish
New Tracking Software
HighWire Press Impact Vizor Citation tracker & distribution surveyor, rejected article & hot article trackers, etc.
Altmetric, Mendeley, CiteULike, etc.
Authorship Criteria
Substantial Contributions to all of the following:
Study conception & design, or acquisition of data, or analysis & interpretation
Drafting & critical revision of the article for important intellectual content
Final approval of the version to be published
Authorship is not justified by:
Participation solely in the collection of data
Acquisition of funding
General supervision of the research group
Each author should have participated sufficiently in the work to take
public responsibility for appropriate portions of the content.
International Council of Medical Journal Editors (http://www.icmje.org/)
Advice
Get statistical help.
Get statistical help before you start study!
Get statistical help as you:
Design the study!
Write the paper!
Respond to the reviewers!
Reviewers
It may help to suggest qualified reviewers
Shows desire to compete at the highest level
Don’t suggest your friends or those with few credentials.
A Pub-Med search of their expertise is only a mouse click away.
Your friends may not be as friendly as you think.
It doesn’t hurt to suggest excluding reviewers with
potential biases or conflicts
But editors may consider these to be precisely the
opinions they want to hear for a fully-informed judgment.
Reviewers’ Common Manuscript Criticisms
Fundamental Problem with the Question
No clear question or explicit hypothesis
Insignificant or irrelevant question
Little or nothing new
Incremental permutation of what’s already known
Poor experimental design & inadequate statistics
Erroneous or unsupported conclusions
Over-interpretation of data
Failure to acknowledge flaws
How To Annoy The Editor Prior or duplicate publications
Non-declaration of work published in languages other than English
Multiple publications from same study
Related data should be reported in one comprehensive manuscript
Several poor publications often replace one good paper
Primary & secondary endpoints are often confused or inflated
Inconsistencies in results are difficult to uncover
Editor’s perspective: another journal gets some, maybe more, credit
Undeclared Conflicts of Interest
Authorship conflicts
Plagiarism
How To Avoid Annoying The Editor
Don’t waste the editor’s time
Follow the Journal’s Instructions To Authors.
Proofread the manuscript before submission
Clear & consistent text & figure legends; data agree with text
Don’t waste journal pages. They cost money. Limit figures, tables, references.
Don’t duplicate data presentation.
Keep the discussion short, focused, balanced Keep interpretations consistent with the data.
Describe how you did or did not answer the question.
Don’t inflate the importance of the findings.
It’s Not Accepted Until It’s Accepted
Letter 1: Accepted
Letter 2: Rejected with minor revisions Generally given opportunity to send in a revision
Letter 3: Rejected with major revisions My be given opportunity to resubmit
Letter 4: Rejected Encouragement to look elsewhere – i.e. really rejected
Letter 5: Rejected “Low priority” – i.e. really rejected, generally for “non scientific”
reasons, inappropriate for journal, etc.
Dealing With Rejection: No, It’s Not Personal.
Almost every published manuscript has been
revised at least once.
Revise before you submit to another journal.
Follow suggestions of reviewers and editors
and try to make most of the changes.
Detail each change & key to original critiques
If you disagree with referee, explain and
support the reasons for disagreement.
AJCN Disclosure Policies
Author Disclosures Signed COI Statement
Almost never disclose non-financial conflicts
Reviewer Disclosures Requested, but voluntary
Trial registration All studies for transparency reasons
Reporting Requirements CONSORT, PRISMA, STROBE, etc.
Disclosure Does Not Correct
Observational Bias Selection and Control Biases Randomization Bias & Concealment Analytical Bias Data Integrity Reporting Bias Publication Bias Allegiance and “White Hat” Biases Fabrication & Falsification
Registration Practices Are Imperfect
Observational studies avoid registration 1° & 2° endpoints rarely specified
Data analysis scheme is rarely presented a priori
Total number of observations are almost never specified
Corrections for testing multiple hypotheses are rarely applied
“p hacking” Investigator Degrees of Freedom are innumerable
RCTs change registration endpoints 5 major journals; 158 RCTs Ewart Ann Fam Med 7:542, 2009
37% changed one or more 1° endpoints
70% changed one or more 2° endpoints
Data are rarely made available for independent analysis & are often not replicable when they are
Research Misconduct
Discriminating unintentional errors from deliberate, intentional fraud can be difficult
From 1977 – 2011:
2047 papers were retracted in the biomedical and life sciences Higher Impact Factor journals were affected most often.
21% claimed unintentional errors
43% were fraud or suspected fraud
The rest were plagiarism & duplication
Müller et al European J Clin Nutr 2014;68:411-415
What is Considered Fraud?
Data Misuse Manipulation of results and/or their presentation
Deliberate concealment of undesired results
Invention and/or submission of false data Disposal of original data
Disturbing or interfering with the research of others
Deception in any form
Participation in the Misconduct of Others Joint knowledge of forgeries of others
Co-authorship of forged publication
Gross neglect of ethical & other responsibilities
Müller et al European J Clin Nutr 2014;68:411-415
Investigation & Consequences
Investigation
Primarily the responsibility of the university
Independent committee or Individual
Report to the U.S. Office of Research Integrity
Or to other appropriate private funding agencies
Consequences
Retraction of paper(s)
Prohibition from submitting grants and papers
Possible dismissal from faculty position
Possible criminal prosecution and prison