Modeling zip+4

What is a "right way" to model (logical model) a postal code, including zip+4? Because the zip+4 was an add-on and is not always used for mailing or analytical purposes, I would think that attributes of postCode (char 6) and zip4 (char 4) might be typical. Would that be the best approach from a theory perspective? Are there either theory or practical advantages to making these either separate or combined attributes? Would it be better as a single char 10 (including -) or char 9 (excluding -) attribute where the first 5 are sometimes extracted for analysis of US data?

I'm doing an illustration for some folks and want show industry best practices (from my perspective, apologies ;-), but I'm bouncing back and forth on this one.

Thanks. --dawn Received on Thu Apr 13 2006 - 17:18:45 CDT