Wojciech,

I would say that publishing of datasets has at least three different purposes:

(1) Accountability; meaning that, given some famous cases of non-existent or adultered data, it seems convenient to show that you have the data, explain how it was generated and prepared, and we are all good (nothing personal).

(2) Reproducibility; meaning that if someone wants to apply the same procedure to your data or evaluate a supposedly better procedure, they can do it. This possibility makes easier evaluating your research and even making contributions. If you do this at the peer-review stage, is even better because referees can solve some doubts looking at your data. Of course, this is even better if you publish your code analysis alongside the dataset; this would save a lot of work to people trying to understand what ou did to the detail.

(3) Reusability; meaning the use of the same data to explore a particular method or even do substantive research with it. Of course, this has to be with corresponding credits to you and other people responsible for the availability of a (hopefully) clean and tidy dataset.

I think that at least the first two are valuable for any data. And even for the third criteria, it may be of use most of the time since, sooner or later, your data could be matched with other small data and form a reasonably big and reasonably representative sample for a new study.In any case, you need to have the corresponding consent from participants and make sure not to disclose any sensitive data.

In any case, you need to have the relevant consent from participants and make sure not to publish any sensitive data.

Hope it helps to make the best decision, cheers!

Manuel Torres-Sahli