Is the rural/suburban/rural thing even based on the pollster's judgment of the poll respondent's address's geography? Or is it based on self-reporting? Do they just ask the person "Do you live in an urban, suburban, or rural area?", and take that as the answer without doing any checking? Do pollsters even ask for the respondent's address in a national poll? If not, then it's got to be based on self-reporting, no?
In theory, it should be known to the pollster, as the pollster knows the phone number they called and can (at least for landlines) associate the first three digits after the area code with a particular geographic area. I.P. Addresses should be somewhat traceable in online polls too - though that might be trickier. The hardest to geographically correlate might be cell phones, since a Nevada resident can have a Los Angeles area code on his cell phone, and cell phone prefixes aren't necessarily as determinative of the exact geographical location where you live, but even then, competent pollsters might be able to use billing info to do so.