10

I'm a bioinformatics bachelor graduate. The group I'm currently working for does not produce its own data. So I often have to search the literature for available data sets. Luckily, tons of interesting data is now freely accessible over the web - which is really great. However, sometimes data is mentioned in the paper, but not published or the download links do not work anymore.

I noticed that asking other groups for data can be really painful. I usually do not get a reply on my first email to the corresponding author. It does not seem to matter how many details I provide in that email about whom I am working for, what I am intending to do with their data, etc. Putting my supervisor in CC does not seem to help either.

The solution is very simple: After a week or so I ask my supervisor to forward my mail to the authors. The next day I usually have a very polite email with the data attached in my mailbox.

This is annoying in many ways: Firstly, I am waiting quite some time for the data. Moreover, I have to ask my supervisor to write emails for me, which is wasting both his and my time.

I'm quite surprised that some (many?) researchers apparently tend to ignore requests from students.

I'm wondering if this is a common problem and I'm looking for advice about how I can improve my emails to increase the chances of getting a reply.

  • 4
    I'm quite surprised that some (many?) researchers apparently tend to ignore requests from students. — Why are you surprised by this? – JeffE Jul 16 '14 at 11:09
  • 3
    @cel Asking for anything that people are not legally required to do sometimes needs social or political power. You have found a way to accomplish your goal, so, what is the problem? If you are afraid of your advisor's time, write e-mail for him (or maybe it is not a big deal for your advisor anyway). – Piotr Migdal Jul 16 '14 at 11:19
  • 3
    @JeffE Do you advocate that it is good that for a given content of an e-mail people should be judged based on their position in hierarchy? – Piotr Migdal Jul 16 '14 at 11:21
  • 3
    I advocate that it is understandable that people reply to other people they actually know before they reply to random strangers. – JeffE Jul 17 '14 at 2:02
  • @JeffE I can understand your point that people might not want to share data with some random student. But I really don't understand why someone would ignore the request of a student of fixing a broken link to data. – cel Jul 19 '14 at 15:38
2

There are similar questions on this site for students requiring source code, such as What action to take when questions regarding a published paper are ignored by its author?. In your case you need data but as you have already seen yourself, sharing data is hardly a volunteer action.

Sharing data is a form of collaboration and collaboration needs equal (or almost equal parties). You benefit from taking the data but how does the other party benefit? You will probably say that the other side will increase his citations but how you guarantee that you will actually get something published if you are just a graduate student at his beginning? When your professor contacts them, two things are guaranteed: a) The confidentiality of the data b) The fair use of the data, including citation. When you contact them, none of these two conditions apply.

Second, many times graduate students just collect data without a clear understanding of what to do with it. It is easier to send a dozen emails to request all sorts of data than actually getting job done on the ones you have. You said it yourself:

...tons of interesting data is now freely accessible over the web

So, why do you need extra data, when there are already many available datasets for you to use? Have you already published something with the freely available data? Have you even compiled a similar dataset? Would you actually share your dataset if you had one? If the answer to most of these questions is NO, then these are the things you must focus on and not waiting to get another dataset for the sake of it.

Also since collaboration requires equal partnership, you will see that these things get easier when you have established yourself, even as a junior researcher. If you have some good publications (using the freely available datasets or the ones that your professor got for you), then when you request additional data (on a polite email introducing yourself and leading to your homepage and google scholar profile), you will not get ignored, because the dataset holder will know two things: a) That you know your craft and his dataset will be used in a fair way (citation included) b) That you may collaborate in the future, so saying NO to you is not beneficial to him. Until then, I do not think there is anything much you can do than using your professor. But since this has already happened more than once (and there are many freely available datasets), I doubt you really need all these datasets you are collecting.

  • 3
    Hi Alexandros, Thanks for your answer. Please don't think of our work as randomly mining datasets for gems and publishing the results. We are trying to answer very specific questions and therefore are looking for datasets from various experiments that we can use for validation and hypothesis testing. It's not like we can just use any dataset, but we're searching the literature for specific data sets for each problem we have. Of course not every dataset will suit. Often you see during the analysis that some question just cannot be answered with the data you have, or your hypothesis is wrong. – cel Jul 16 '14 at 10:08
  • 1
    @cel "...I should have access to all data that is used in some figure to see if their claims are right". No, this is wrong. Their work has already passed peer-review. You are not their reviewer, nor do they have to prove anything specifically to you. – Alexandros Jul 16 '14 at 10:25
  • 7
    But in my opinion that's not what science is supposed to be. Things are not right just because you found some other scientist who believes you. If you published your hypothesis you should expect that the scientific community would like to check your results as well. But that's more philosophically than helping in this argument, I guess. – cel Jul 16 '14 at 10:43
  • 2
    @mmh I have no tools to check whether you methods are bad. I am just saying that studies that were not made as much reproducible don't push science forward (as far as they could). Sadly, I am aware that the current academic system rewarding only papers is suboptimal. At least, if there is embargo period to write a few papers, do you intend to publish (i.e. make public) the data one day (I guess you don't need to pay for doing so). – Piotr Migdal Jul 16 '14 at 13:17
  • 4
    You could read the methods section from my paper. The costs for publishing the data would be: 1) Because it is possible to recognize and individual from an MRI, I would need to anonymize the images and the associated metadata, 2) I would need to ask permission from an ethical committee, 3) I would need a significant amount of disk space for hosting the data. I'd say the overall cost is > 1k€. Anyway, I do agree with you that there would be benefits from open data. However, it does not come for free. – mmh Jul 16 '14 at 14:06

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.