Problems with Writing a Paper
There are also problems associated with the preparation and submission of manuscripts. These problems are ethical problems, not those associated with writer's block or similar difficulties. Using other people's words or data (plagiarism). It seems unnecessary to discuss the ethical problems associated with "stealing" someone's property, but such theft is common. Every school child hears (or should hear), at some time during his career, an admonition against plagiarism. It's hard to believe that any person who is native to American society does not know the societal admonitions against it. Yet, nearly every year there is someone in my Scientific Writing course who must be reminded that "another author said it so much better" is not an excuse for stealing his words without giving him credit. It's almost impossible to read about a problem and not incorporate into one's own writing some words or phrases taken from something you've read. The writer may not even be aware that he has done it. After all, we learn English in the first place by imitating what we hear. That is quite a different matter from knowingly stealing someone else's words. The infraction is only slightly less severe if the other person's words are only paraphrased, rewritten slightly to avoid an exact duplication. Stealing data is a more serious matter. This kind of theft can never be done unconsciously; the perpetrator always realizes (or, at least, he should) that what he is doing is wrong. The consequences of this act should be the same as the consequences for making up data. Not reporting contradictory observations you made. There are times when the results of repetition of an experiment may be different from the original results because of chance or because of some critical change, even a small one, in the procedures. Sometimes control values may be different from one experiment to the next, sometimes experimental values. If you notice this kind of discrepancy, it is unethical not to report it in print. Putting your name on work you didn't do. There are clearly strong pressures for a scientist's name to appear on as many publications as possible. These have led to an explosion of author lists on papers and to inclusion in such lists of people who made no substantial contribution to the work or know nothing about it. It is not always clear whose name should be put on a paper. There are certainly people who have a legitimate claim to authorship, but there are many who do not. Recent fraud cases involving "innocent" co-authors, who did not know that the misconduct was occurring, show clearly the problems with including individuals who had no substantial involvement in the research. If they were not close enough to the research to know the fraud was occurring, they should not have been co-authors! The problem of authorship is acute for students because they may feel, wrongly, that they do not have the "stature" to make judgments. The professor may claim a right to authorship because he supplied the funds under which the work was done or the laboratory space required. This reason alone is probably not sufficient. On the other hand, if the work was a specific part of the grant proposal that won the money in the first place, the claim may be justified. Many professors will leave it to the student to decide whether the professor's contribution was sufficient to merit authorship. Inclusion of others beside the professor as authors may be a thorny problem. Fahmy and Fahmy (1991) suggest that authorship be restricted to "those who designed the study, supervised the process of data collection, and are responsible for the content of the manuscript." They claim that this will limit author lists to one to three individuals. Even so, this still leaves open the question of the order of the remaining authors (Riessenberg & Lundberg, 1990). Not reporting others' related or contradictory work. Eugene Garfield (1982) has pointed out that citing other scientists' work is part of the reward system of science.
"After all, citations are the reward system of scientific publication. To cite someone is to acknowledge that person's impact on subsequent work. Citations are the currency by which we repay the intellectual debt we owe our predecessors. Furthermore, failing to cite sources deprives other researchers of the information contained in those sources, and may lead to duplication of effort." Failing to cite the work of others, work that is related and complementary to your own, is, as Garfield points out, bad form; it's depriving others of their reward. Still, it can be difficult to decide what to reference. Working in a popular field, there may be a great many previous papers on the topic. It may be impossible to cite them all. There is no set formula for this decision; you simply decide which are most important to the case you wish to make. Those whose work you miss may have their feelings hurt. Perhaps a statement that there is other work uncited will suffice. Garfield asserts that for "any scientific paper, there are certain earlier works that must be cited. These are the papers that any honest scholar would be ashamed to omit from his bibliography--and that any careful referee would insist on." Clearly, there is another group of earlier works whose importance to the paper may be debated by different referees. More important to the progress of science is the failure to cite work that is contradictory. It is not unusual to find the members of a camp supporting one point of view on a scientific issue not citing work by members of an opposing camp. Such pettiness has no place in science! A scientist owes it to his science and to his reader to discuss all relevant aspects of the problem he has studied, including contradictory points of view. If he himself stands on firm ground, he should have no difficulty dealing in print with his opposition. Changing the hypothesis for the paper. Scientists are probably good guessers. That is, they probably, more often than not, form hypotheses that are correct. But, it is unlikely that the scientific community guesses correctly more than 90% of the time. Yet, a casual glance at the literature shows that the number of failures to confirm the stated hypothesis of the experiment is small, much smaller than you would predict. There are probably at least two reasons for this small number. First, authors dislike reporting "negative results," and editors dislike publishing them. Therefore, such negative results are sometimes not submitted for publication, and, if they are, they are not accepted and published. This is unfortunate for the progress of science, because there are probably many unnecessary replications of work that produced negative results, replications that would have been prevented if such results appeared in the literature. Another reason for the low number of failures to confirm hypotheses is that the hypothesis stated in the paper is the converse of the one actually tested. In other words, the investigator claims that he obtained results supporting one hypothesis when actually he obtained results inconsistent with the original, opposite hypothesis! What, you ask, is wrong with that? It is a clear violation of the rules of evidence. An investigator may not use data that suggested an hypothesis (in this case the converse of the one actually tested) to support that hypothesis. Only data collected to test an hypothesis may be used to support it. In this situation, another experiment must be done to obtain data supporting or refuting the converse hypothesis. Gradually changing from "far out possibility" to "established fact." This is an insidious problem. A paper may begin with a statement that a given hypothesis has a low probability of being correct. As the paper develops, the hypothesis gradually becomes "likely," then "certain." The abstract of the paper states only that the author has "shown" the hypothesis to be true, when, in fact, he has not even tested the truth of the hypothesis. He may not even have suggested alternative hypotheses to be contrasted with his own. Believe-it-or-not there are readers who read only abstracts, and they would be totally deceived by this paper. Readers must be careful not to fall into this trap. Authors should be careful and honest in not leading a reader into the trap in the first place. Concluding "cause and effect" when only "correlation" is demonstrated. The brightness of day, as compared with night, is caused by the relative position of the sun above the position on earth of the observer. The symptoms of Parkinson's disease are correlated with the absence of dopamine from the substantia nigra. In the former case, causation is known; in the latter it is not known. Far too often investigators behave as if they don't know the difference. The reader of the paper should know the difference and not be fooled by the incorrect assertion of causation, but not all readers read carefully and many students may read the paper, not knowing the difference. In any case, ethical behavior demands that the difference between causation and correlation be acknowledged and that authors indicate when only correlation can be claimed. Writing an abstract with no data. Sometimes an abstract is written before any data were collected in the experiments it describes. In fact, careful reading of the abstract shows that it describes the procedure but says nothing about results. With deadlines for submission of abstracts six months or more in advance of the meeting, it is not unusual for authors to presage their anticipated results in abstracts. In one case that I know, when the meeting actually occurred, the author of the abstract delivered a talk that described the procedure and gave no results. The reason: there were no results to give. The author never said that he could not record from any cells in the brain in his heart-lung preparation, and the abstract was never withdrawn as it should have been. That clearly is unethical. Failing to report negative results. This is a problem similar to editing data. All of the results of an experiment but one can be similar. That one could have been deviant for a number of different reasons. Perhaps the animal or human from which it was obtained was sick or otherwise abnormal. Perhaps the equipment was not working well that day or a particular assay may have gone wrong. The experimenter recording the data could have written the values incorrectly in the data book, perhaps transposing digits. Any of these explanations and possibly others could be used as arguments for discarding that one measurement. On the other hand, it could be that the parameter being measured is highly variable, and, by chance, the data were selected from the "middle" of the distribution where the values are close together, the one deviant value being selected from the more distant "tail" of the distribution. One rarely is able to apprehend the reason for a deviant value. What should be done? There are a number of approaches that could be taken. The first is to include the value with all of the others. If the results are insignificant because of that value, then a larger sample will have to be drawn. The statistical evaluations could also be made with and without that value and the results of both analyses presented along with an argument as to why the deviant value should be disregarded. In this case, it is left to the reader to decide whether he believes the value belongs in the sample or not. Either approach would be valid; the latter makes more thorough use of the data. Publishing the same results many times. This problem shows up in several different ways. First, some authors have published exactly the same paper in two different journals. Whereas this practice is not as common as it once was, it does happen, and one occurrence is too many (See Lock, 1984). Another form of this problem is the publication in a journal and then in multiple review articles of identical data. Albeit there is no rule against this, it does get a bit dull reading the same thing multiple times. The third form of this problem is what Lock (1984) calls the "salami" series of articles in which the data are published piecemeal instead of all together. Some authors publish the report of a project on the installment plan. They report the results of a small group of subjects, then, those of a larger group and finally the entire study. Still others do a study, parcel it into small units and then publish each small part as a separate paper. Again, this is not prohibited, but it is not a good practice to pad one's curriculum vita while forcing the reader to synthesize what the author should have done. Failing to allow review of the manuscript or the obtain approval from all authors. Every author whose name will appear on a published paper has the right and the responsibility to review the paper before it is submitted. He should know what the paper is about, understand the rationale for the experiments, understand the methods used, know the results obtained, and know the conclusions drawn. Furthermore, he should agree completely with everything in the paper. Anyone who has not done all of these, should not appear in the author list. It may be surprising, but there are authors who did not know their names were included on a paper until they encountered the paper while browsing in the journal. Sometimes they have been totally shocked by the viewpoints appearing in print under their names, viewpoints with which they could not agree. Just as it is unethical to leave a deserving coauthor out of the author list, it is unethical to include an undeserving or uninformed coauthor. Where to Go From Here:Introduction
|