Let's say David Smith has published a paper a while ago. In this paper there are some results plotted in a graph. I want to use this dataset in my own research and extend it with my own results. Unfortunately, it is not possible to contact David Smith. So I use some of tools mentioned for example in this question to extract the dataset from the plot (which, of course, might introduce some kind of small errors).

Do you consider it as OK if I now publish a paper which includes the old dataset from David Smith + my new data in a new plot? I would, of course, cite the old paper by David Smith. (I might deduce some results from the old+new data that a somewhat different that the original conclusion by David Smith.)

  • Can you elaborate why you have doubts regarding if it is OK? – Roland May 17 '17 at 10:23
  • @Roland my doubts are about extracting the data with software-tools from their plots and do further research with it - instead of having the proper (real/raw) dataset and allowance from the original author. – Alf May 17 '17 at 10:27
  • I'd understood what your doubts are about. What I'm interested in is why you have doubts. Such information could help writing more specific answers. Extracting data from graphs has a long tradition. I can't see how anybody would suspect ethical problems if proper citation rules are observed. That's why I'm asking you for your reasoning. – Roland May 17 '17 at 10:41
  • @Roland I see, let my try to explain: I think my personal feeling is that using these tools to extract data feels a bit like just copying somebody's data + I might introduce errors by extracting the data which could lead to wrong conclusions about the original dataset (only on a first check, though, a through check would/should reveal that I made an error). But that issue is tackled by JayFromA's answer. – Alf May 17 '17 at 10:46
  • Extracting data shouldn't introduce errors if done right. It usually introduces additional uncertainty which can be taken into account for subsequent analyses. A graph is just a different way of presenting information than a table. Would you have the same doubts with extracting data from a table? – Roland May 17 '17 at 10:49

If you cite the original authors in your plots and when comparing results/conclusions (or simply everywhere you use their data/work) you should be fine. Also make it clear that the data is "reverse engineered" from their graphs/plots and not the "original" data.

  • 2
    clearly stating that it is "reversed engineered" instead of saying "old data from Ref. [xx]", I like that idea, thanks! – Alf May 17 '17 at 10:28
  • 1
    Good answer. I would also like to mention that meta-analyses are doing exactly that, with lots of articles! – Emilie May 17 '17 at 12:47

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.