Press: Teacher Li Honghan's "is this data fraud?" The article mentioned and discussed the problem of how to treat abnormal data in scientific research. Although this problem is a platitude, it is indeed very important. At any time and at any level, scientists should always attach great importance to this problem. The following is the author's detailed comments on this problem.

First of all, "when can abnormal data and abnormal phenomena in scientific research be eliminated, when can they not be eliminated, under what circumstances the elimination is scientific and reasonable, and under what circumstances the elimination is unscientific and reasonable?" there are thinking traps and psychological traps that are very easy to fall into. Both new students and successful scholars need to always be correct We should attach great importance to this problem and be self-conscious, because once we fall into this trap (those that should not be eliminated are easily eliminated) and can not self-examine and self correct, it means that (1) the parties themselves have no real access to scientific research, or it means that (2) they are beginning to form serious thinking patterns and thinking rigidities (or they are too arrogant to know themselves) ）As a result, the ability of scientific research has been on the decline. Either of the above two situations requires self-awareness and self correction as much as possible.

Secondly, the existence of abnormal data should have been a great opportunity to improve scientific theory and even make scientific discoveries. If we deliberately and selectively look for evidence to prove our own point of view, it is not only a wrong research method, but also a kind of academic misconduct or even academic fraud (if we deliberately eliminate abnormal data cases without scientific basis) It's too late. The real scientific approach should be to deeply study the root of abnormal data and abnormal cases, and try to keep all normal data and abnormal data as completely as possible. At the same time, we should deeply study whether some samples or data results in some cases are abnormal (this situation is an opportunity, which should be cherished, because this may be an excellent opportunity to make scientific discoveries), or those abnormal data The data is caused by mistakes and errors in experimental operation or experimental design (this situation is an opportunity for self-examination and self-improvement, which indicates that there are deficiencies in their experimental skills and scientific literacy, and they need to improve their scientific skills and scientific literacy). In short, no matter whether there are abnormal data and abnormal cases, in the process of experiment, there must be parallel sample and contrast experiment, there must be repeated test experiment process, and the experiment design, experiment operation and data collection should be carried out as completely as possible (at least with enough representativeness). This kind of repeated test and repeated test should not be carried out on the research data, so it is easy to eliminate them Abnormal experimental data and abnormal cases are nonstandard and wrong research methods.