Ask. Then collect.
Any data that people say is fine to use - and that they fully understand the use of, likely potential risks (to the best of a project's capabilities to consider) - should be considered fair game.
Anything else should not.
People's identities are sacrosanct. On computer systems, these identities are data sets - and it is the people they represent who should have full control of them.
If that is not possible, do not collect the data.
* Check out the post linked for a little more discussion of one sort of design (certainly not the only ; - )) that might help balance planning wants with individual rights (and cover planners' butts in so doing ; - )) *
- More Info:
- Build the foundation, then build the rest