Is validating the results of an article important before citing such an article?

Question

During writing my academic paper, I need to cite accuracy values of cited research paper.

I have been advised that if you will cite a paper, you need to read that paper entirely. But in this case, Do i need to validate results that they achieved.

Answer 1

Reading yes. It is important to avoid extraction of sentences or results which can be less general than as they appear at the first glance.

Validating not at all, unless you feel you must do it because something is suspicious or you think to deal with a particular or new case.

You can't validate each statement for every cited reference, and you can't do so for several reasons. Also a referee does not. Papers exists to be used. That is an implicit form of validation.

Answer 2

This depends on the situation. In some cases it is impossible to do so, not having access to all the necessary materials. If you can do so with reasonable effort, however, it is good to do this, just to satisfy yourself.

But, another consideration is the quality of the source/journal. If the journal is of high quality with a good reputation, then you can probably depend on the journals review and editorial process to give assurance to the results of the paper. It isn't a perfect process, of course, but it adds some evidence of validity at least.

But, it is poor practice to cite things without reading them, as you say. It is also poor practice to cite things that you either don't understand or believe to be incorrect. The first requires more work, and the latter may require more research and a correction.

Answer 3

The idea of peer review is to remove this element of research. If you had to validate all previous work you are building upon you would need to start science from scratch each time...

If the work is published in a reputable journal - then validation is absolutely not necessary. However, you should always read the work you are citing!

If the paper is not published in a reputable journal and you are dubious about the results, then either find a higher quality citation or try to replicate the results (if possible).

Answer 4

You should do enough work to convince yourself that the paper you cite is valid (or that it's invalid, if you're citing them to critique them). So that means:

• Reading enough of the paper that you are sure you understand the part of it that you're referring to when you cite it. If you're using paragraph 4.2, make sure you know what section 4 as a whole is about for example. You don't want to be surprised if someone points out that paragraph 4.4 contains some limits on what 4.2 does.

  It's probably safest to read the whole paper, but you don't have to read equally "deep" in the parts that aren't really related to what you're doing. But make sure you read the introduction, look through the related work, and the conclusions.

• To convince yourself that the paper is valid, you look at several different things that can show you that there are problems or that can show that it's high quality:

  • Is the paper written clearly, with the normal structure you would expect for a paper in your academic field?
  • Are experiments clearly described? Does the methodology seem reasonable?
  • Are the results properly described?
  • Do the conclusions the paper draws make sense, given the results?
  • Is the paper written by someone with a good reputation, someone with experience in that field? What was their academic title when they wrote it?
  • Is the paper published in a journal with a good reputation/peer reviewed?
  • Is the paper cited by others? What do people citing the paper (in Related Work sections for example) have to say about it?
  • Look at the References section of the paper; who does it cite? A possible sign of a bad paper is that it doesn't refer to the normal literature for that topic. That could mean it's a crackpot working on crazy theories and disregarding normal science in the field.

Doing these things, you can get a good impression of whether a paper could be valid or not, without having to re-do all their experiments.