5

Within fields like animal behavior, data is often collected through observation of the subjects, either in the wild or captivity.

  • Can a paper be published purely based on observational data, with no evidence in video or audio to support it?
  • If so, could one (in theory) generate an entirely fictional dataset of observations as basis for a publication? Or do other mechanisms exist to prevent this kind of fraude from happening?
  • Are there examples of publications, in animal behavior or other fields, being exposed for generating completely fictional datasets? If so, which?
  • 1
    Can a paper be published purely based on observational data, with no evidence in video or audio to support it? Maybe search the literature. If so, could one (in theory) generate an entirely fictional dataset of observations as basis for a publication? Yes, that follows immediately from your premise. Are there examples of publications, in animal behavior or other fields, being exposed for generating completely fictional datasets? Yes, such examples exist. – user2768 Jul 22 at 6:59
  • 9
    Even video/audio might not be sufficient proof. Assume that Observation X was made 10 times when Y was present. However, the author deleted the 90 cases where Observation X was NOT observed when Y was present. – J-Kun Jul 22 at 8:30
  • 3
    You should take a research methods class. It will teach you how best to make observations so that they are scientifically valid. Such a class will also teach you what the standard is for publishing data, and what kind of corroboration is necessary. – Wolfgang Bangerth Jul 22 at 11:26
  • 2
    Some of these comments are really answers. – Buffy Jul 22 at 11:53
  • 2
    Wikipedia has a (no doubt partial) list of scientific misconduct incidents. Some of these are just plagiarism cases, but many involve falsification and/or selective presentation of data. – Michael Seifert Jul 22 at 18:11
12

In many experimental fields, it is possible to "generate" a plausible amount of data and get away with it (i.e. publish the data and the conclusions in a peer reviewed journal).

Experiments need to be repeated independently. Otherwise, all sorts of things happen. Real fraud is probably the lowest number. Much more important are cases in which experiments are done without enough care or statistical methods are used in a questionable way.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.