Workshop | Topics: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | Other | References | Mailing list | Welcome

Topic 8. Validity and scoring methods.

What makes results useful? Do the results have to be scored by the data holder or participants contacted? What about merely reporting a statistical probability of being correct? What about just providing a proof of concept or reasonable argument of vulnerability? What is the responsibility or vulnerability of those providing the data?

What are important issues? What are risks and harms? Which issues are most likely to occur, and if they do occur, which are most likely to have significant adverse impact?


Post 1
The results are useful if they teach you (with prove-able results) something that you
were not aware of before, or proving something you already knew. The results do not
need to be scored if there is a possibility that the results can be corroborate by
someone at some other point in time. Just providing a proof of concept or reasonable
argument are useful pieces of information, and simply reporting a statistical probability
is extremely useful in many situations.

Post 2
Results are useful if they teach you something you didn't know before, or
confirm something you suspected. Results don't need to be scored by
participants contacted. Reporting a statistical probability of being correct is
not only sufficient but very useful.

Post 3
The results are useful if it expands a current knowledge base or confirms a previous study.

BOOM.

Post 4
It depends on the stage of the debate in re-identification. It seems that the situation
currently requires more proof of concept and persuasive arguments for vulnerability,
before work and policy can advance.

Post 5
Results do not need to be scored by participants or the data holder. If they are scored
by the data holder, this can be seen as a further invasion of privacy, because the
identification goes from a probabilistic assessment to a certainty.

What makes results useful depends on the researcher's goals in reidentification. (i.e.,
Why are they trying to reidentify people at all?)



IQSS  |    Data Privacy Lab  |    Silent Spring Institute   |    Northeastern University