There are limits to preserving research data: there is neither sufficient money nor space to keep everything. On what basis do you decide which data must be preserved? And what does it take to destroy data?
When selecting research data for long-term storage, the first thing to consider is what the requirements of your research funder or publisher are. Chances are that you are obliged to keep your data for a specific period of time, with a view to their possible reuse or for checking your research findings.
In compliance with the Netherlands Code of Conduct for Research Integrity, UvA and AUAS request researchers to keep their raw research data for a suitable period. A commonly used period is a minimum of ten years.
The second point to consider is the importance of your research data to the discipline and possibly to other disciplines, for example, data that are very likely to be reused and data from highly innovative research.
The third point to consider is the value of your data. Data with non-repeatable observations are unique and cannot be created again by others. If the cost of (re)acquiring the data is very high, this might also be an argument to preserve the data.
Secondary selection criteria are the usability of the data to be kept: is the file format (still) usable? Are the data accompanied by the appropriate documentation and metadata, so that it is clear how they were acquired and what they mean? Is it clear who has the intellectual property rights?
Data files which are not preserved, must be destroyed carefully to prevent the data from getting into the hands of unauthorised persons.
The safest way to destroy data is to make the medium (hard disk, USB stick, DVD) completely unusable physically.
It is not sufficient to remove the file from the hard disk, for then it is still possible to recover the file. (Deleted) files must be overwritten to make them inaccessible. This can be done by means of software.
Please note: if you use Windows, be sure to install CCleaner version 5.34 or higher (more information).