Should I (student) share my data with a researcher I don't know directly?

Question

I am currently a graduate student, and for a course I gathered hundreds of thousands of records (not confidential, but difficult to access if you don't already know about them and know who to talk to) and spent several months cleaning, combining, and organizing them into a usable dataset, upon which I then performed statistical analysis. The project is complete and produced interesting results, but likely I won't turn it into a paper anytime soon.

I found out yesterday that the professor supervising the course spoke to one of his friends and mentioned my project, and the friend asked for my dataset (the cleaned one I produced, not the raw records) to perform his own analysis. Should I share it with this friend? If so, is there a way to ask to be acknowledged in any publications that result?

The data were originally public records, but I did a lot of work that required years of specialized subject-matter knowledge to compile them appropriately. Are there other risks I haven't considered? I feel a little uncomfortable being asked to share a large amount of work with an academic I don't know at all, and while I would like to help advance the field in general, I don't know what's reasonable to expect here.

Update As suggested here, I emailed the advisor saying I was not comfortable giving the data to someone I did not know (he wanted me to upload it to a file system I was shortly to lose access to), but that I'd be happy to discuss with his friend directly if he wanted to put us in contact. He responded saying that he would look for a different dataset.

Answer 1

TL;DR - talk to your supervisor (which to be honest is the answer to about 90% of the questions on here, but hopefully this answer helps structure that discussion).

First of all, you might need to check whether you are even in a position to make this decision. The ownership of intellectual property arising from student projects can be a complex subject and you might want to check with your Knowledge Exchange and Commercialisation office or equivalent before sending a large dataset to an external partner. In most cases this is likely to be a formality and there will be no issues unless the dataset might be used commercially, but they may want to put a simple Material Transfer Agreement in place to define how the recipient can use or further distribute the dataset.

Assuming it's up to you, essentially you have three options:

1. give them the data without restriction. Best-case scenario: co-authorship, worst-case: nothing.
2. refuse to give them the data. Best case scenario: nothing. Worst-case scenario: you and your supervisor look bad and you damage a potential future relationship with a collaborator or employer.
3. give them the data after taking some actions to increase the likelihood of acknowledgement. There are three real options here:

3a) (mentioned by @gvgramazio) upload the data to a public repository. The preferred repository varies by field but I'd use an academic data repository like Figshare. You get a citeable DOI but this is not a publication, so if you're looking for publications, this doesn't help (it should, but that's another discussion).

3b) publish the dataset in a data journal like Nature Scientific Data; then everyone can use it but has to acknowledge you by citing the source.

3c) set up a material transfer agreement that defines how the recipient can use the data and what you get in return, which can formally include the option of coauthorship in some form if a manuscript is produced.

With option 1, the most likely outcome depends on their character. Are they a good person to collaborate with? Do they have a good reputation as a collaborator? This is probably the most important question, since if they're a good person to work with then I honestly can't see them not including you as a co-author anyway in this situation, although this could vary by field. Another, related consideration is whether might you potentially be working with (or for) them in the future, and if so if this might increase your visibility or chance of a job. If you don't know them, the only real way to assess this is by asking your supervisor, who knows them and also has an interest in supporting you.

Option 2 - refusing outright - really isn't likely to end well.

Option (3a) - a data repository - guarantees acknowledgement but not necessarily in a form that's useful to you, and option (3b) - a data paper - is the most work and guarantees acknowledgement in a form that's likely to be useful to you, but both (3a) and (3b) mean you're less likely to get a co-authorship.

Option (3c) is trickier. In my field MTAs are increasingly common but also very unpopular with some researchers. As the provider of the data, having it covered by an MTA is the safest option but negotiating one can be intimidating. Fortunately, most institutions have offices to handle this sort of thing.

You also need to consider a few other questions:

• how helpful are the different possible outcomes going to be for your career? In my field you are assessed primarily on publications in peer-reviewed journals so 3b is preferable to 3a. However, if the friend is able to acknowledge the data by citing it, they are less likely to include you as a co-author unless you make further substantial contributions to the analysis (which they don't have to give you the opportunity to do).
• Does their proposed analysis sound exciting? If it has a good chance of resulting in a really, really good paper and relies heavily on your dataset, then a chance of a co-authorship on a really high impact paper might be better than a lead authorship on a data-only paper.

Ultimately, most of these questions will be things your supervisor is able to help you with. So I'd arrange to talk to your supervisor, explain your concerns, and ask which option they suggest.

Answer 2

Why would you not want to share the data you have collected and prepared, collaborate with an experienced scientist, and finally make a paper from months of work you put into this project already?

I would rather see a good opportunity than a risk here. The highest risk is that someone else receives all the credit for your hard work. Well, while that is certainly not nice, you could probably live with it, especially when you are not interested in publishing the work on your own anyway.

However, it is much more likely that the fellow researcher will respect your work and give appropriate credit (like a co-authorship). Make sure to send a description of the data preparation you have done along with the data, so he is aware of the amount of work it has taken. You should also show clear interest in writing the corresponding sections of the paper (or whatever publication is planned).

Collaboration is one of the key ingredients to excellent research. Go ahead!

Answer 3

Your decision is based on:

1. Is it ok for you to share your work, in general?
2. Is it ok for you to share your work and be acknowledged for that?
3. Would you like to work with your advisor's friend?
4. Do you have any personal grudge against your advisor?

If it's not ok for you to share your work or if you have some personal grudge against your advisor, then simply don't give your data to them. You could find any excuse in order to not do so.

If you would like to join in a project with your advisor's friend then you could exchange your data for it.

Finally, and I think this is your case, if your only problem is that you want to be acknowledged for your efforts but you don't have the opportunity to publish anything soon then simply publish your data under license. You could do it on GitLab, GitHub, Bitbucket, etc. You don't need to make your repo public, the important thing is that you put your license. Then you can give your advisor's friend access to your data and he's obliged to acknowledge you in his papers or you can sue him.