One key aspect in inspiring confidence in the use of the data is to be accountable for what happens to it. This means at the end of the study, the data should be destroyed. An initial reaction is that the Evaluator just has to delete the data files. But is it that simple?
Actually, there are several points to be considered and although not complete, this article gives the general flavour. First, what actually happens when a file is deleted? Actually, no more than is what is obvious. The file no longer appears on the list of files for the directory. A skillful hacker who can get around the computer's operating system will find the data on the hard disk untouched. Technically, the data would remain on the disk until another file is written over the data.
The solution to this will depend on the version of the operating system that you have installed. All mature operating systems will have some variant on the traditional delete function that will overwrite the existing data so that it no longer exists. Still this is not enough for some. Although I know of no actual cases in program evaluation, it is considered a real possibility that when the data is being overwritten that old data will be missed and still be present on the disk. Commercial products exist to perform a very thorough wipe of a disk to eliminate the possibility.
It is just not the operating system that may leave stray copies of the data, the statistical software may create data in intermediate forms that may be in unusual formats. R is a good example of this as analysts are encouraged to collate the data in refined file formats, which may be more revealing of the personal data than the original.
There is another potential danger that is very real in the world of program evaluation when reproducibility of results is an issue. This derives from the necessity to be able to demonstrate how results are achieved and that they are valid. There is no way to achieve this without maintaining the data for certain period as specified in the privacy agreements. However to be able to achieve this, the organization that the evaluator works in must have an effective form of archive management in place, so that the organization can be assured that the data is deleted, even if the evaluator has left the job.
What is the moral of the story? Before you request data, make sure you have a plan in place. Data management is not as simply as it looks.
| Hard Drive being physically destroyed |