Just because many people
chose to model surveys and
questionnaires as flat tables
doesn't mean it's correct!
Again, because you know absolutely nothing about her database objective,
doesn't automatically mean that "not properly normalized", as you implied,
was the case either. My point was that without knowing anything about her
database, you SHOULD NOT imply that it's "not properly normalized" because
it might be a survey, questionnaire, etc, which ended up being the case.
FYI: Simple questionnaires are usually sent out to get answers from a SINGLE
ENTITY and that's it! Then those answers come back and get sucked in by
another larger process, which is where the relational model usually comes
into play. Note that I used "simple questionnaires" [which is typical
practice]. Meaning, if I'm getting information from Fidelity about the
number of years they have been in business, the number of customers they
have, their customer attrition percentage since inception, their customer
retention percentage since inception, etc, then there is no need to have
anything other than a flat table. And surely a questionnaire would not ask
for their customer demographic information. I won't contribute to this
thread turning into a debate on whether a questionnaire should be relational
or not. Again, my point was that without knowing anything about her
database, you SHOULD NOT imply that it's "not properly normalized" because
it might be a survey, questionnaire, etc, which ended up being the case.
It's safer to simply answer the question without trying to blindly dissect
someone's database design just because YOU can't imagine a design like hers.