Former IMS Bulletin Editor, and current Co-editor of JASA, Xuming He, writes:
In order to disseminate our research results, we are expected to publish. We also publish to establish a track record as scholars. Publications are needed for annual evaluations as well as tenure and promotion for each individual. Publications also play a role in the rankings of programs and institutions. According to the NSF report Science and Engineering Indicators 2012, the number of research articles published in a set of international, peer-reviewed journals has grown from about 460,200 in 1988 to an estimated 788,300 in 2009. The trend is probably continuing at an even faster pace today.
When we as authors are under pressure to write and to publish, it is important that we do not forget about the basic principles of research ethics. I am no authority in this area and do not intend to present myself as one. I will however bring to your attention a few examples that I have seen as editor and reviewer, hoping that we can all avoid the temptation of taking an easy route to publications.
Example 1: Simultaneous submissions of multiple manuscripts with substantial overlap.
We all know that simultaneous submission of the same manuscript to more than one journal is prohibited. But papers that develop the same ideas in different problems or applications do get published. It is important that you cite the other papers that use the same ideas and discuss the connections and differences between your current paper and the earlier ones. If you submit two papers around the same time that apply the same idea to two different settings but do not cross reference each other, you might be viewed as dishonest.
Example 2: Fabrication or falsifications?
We all know that fabrication and falsifications are unacceptable in research. But the same applies to various forms of dishonest and selective reporting. If a simulation experiment requires a tuning parameter, and you report only the results that are favorable to you without specifying how your tuning is performed, you could be cheating. It is fine to demonstrate the superior performance of your favorite method for well-specified settings, but make sure you keep all the computer code that is used to produce your results. You should make them available to other researchers too.
Example 3: Plagiarism.
Non-native English speakers (and many others too) may find it especially tempting to use well-written statements or paragraphs from the existing literature in their own papers. This however may lead to plagiarism. Self-plagiarism, that is, repeating your own writing without citation, is harder to detect, but it is not appropriate either. To avoiding plagiarism, you should consider using quotation marks around material taken verbatim from a source or using different words to summarize what you have learned from the literature. Information that you obtained from a private conversation, correspondence, or discussion with third parties should not be used in your paper without permission.
Example 4: Evasion of responsibility.
When asked about a possible error or confusion in a publication, some authors simply point to their coauthors for responsibility. “This part of the proof was provided by my coauthor” or “The computation was carried out by my coauthor” might be honest statements, but they do not relieve you from responsibility. If you are a coauthor, you must approve the paper as a whole, and consequently take responsibility of what you say in the paper. There are cases, however, where you are responsible for the statistical analysis of a paper but the rest of the work falls outside your areas of expertise. In those cases, it helps to state in the paper who are primarily responsible for each part of the paper.
These examples show that we authors have to think carefully about what we publish. The pressure that stems from the “publish or perish” mentality might be there, and the temptations of easier and faster publication confronts many of us. However, we should remind ourselves that ultimately we will be judged not by how many papers we publish, but by what we publish.