In two previous posts, I have mentioned two of the seven sins of academics in the natural sciences mentioned in an article by van Gunsteren (Angew. Chem. Int Ed. 52: 118 – 122). The third sin was insufficient connection between data and hypothesis, or over-interpretation of data. My personal view is this is not a sin at all, as long as you are honest about what you are doing. Perhaps the best-known example is that of Kepler. Strictly speaking, his data were not really sufficiently robust to justify his laws, but Kepler decided (correctly) that the planets should follow some sort of function, and the ellipse fitted the data better than anything else. Similarly, in one sense it was an act of faith for Newton to accept Kepler's laws as laws, but look what came from it. My view is, that as long as you are honest, there is no harm in drawing a conclusion from data that does not fully support it, as long as it is clear what you are doing, and as long as the conclusion is not put to a critical use. This if considering whether something is safe, then if the data does not prove safety, it does not hurt to hypothesise that it could be safe, as the hypothesis takes everyone forward, but only if it is clear that it is a hypothesis.
The next sin mentioned is the reporting of only favourable results. Here I am in total agreement. If some result does not support your hypothesis, you should investigate it thoroughly, and if it persists, you not only report it, but you should confess that the hypothesis is wrong as stated. To me, it is a sin, albeit a less serious one, to report the data and make no comment on it. The statement that it is unexpected, or to state it and end the sentence with an exclamation mark, is not adequate. The reason for this is that in logic, ONE observation that cannot be explained by the theory is sufficient to falsify it.
Another sin mentioned was the neglect of errors found after publication. If the error is in the reporting of the data, such as a spectral peak listed in the wrong place, obviously this should be reported. However, I am less sure of reporting when the report does not make a significant difference and does not conclude the matter. In my opinion, it is almost as big a sin to put out a sequence of papers on the same subject, and having a conclusion that moves around a little from paper to paper. If the first conclusion is near enough, in my opinion there should be no corrections until the author is convinced the subject is sorted. There is far too much in the literature already, without salting and peppering it with minor variations, none of which significantly improve the issue.
The remaining sins listed were plagiarism and the direct fabrication of data. I agree these are bad sins, but do they actually happen? I have heard there are examples from students, but surely this is as much the fault of supervisors. I would hope that professional scientists would never even think of this. As far as I know, I have never run across an example of either of these. Have you?
I realize these opinions might be controversial, but so what? I hope it does stimulate discussion. I also think the list given in this article is incomplete, and I feel there are more sins that are equally bad (except possibly for the last two). More on them some other time.
