The question arises because of reports in the media about the increasing number of households-particularly those with younger adults-who have chosen not to install land-line phones, relying instead on their cell phones. Some have wondered whether it is still possible to get a representative sample of adults or likely voters nationwide, knowing that an increasing number of younger voters use only cell phones.

The answer to that question is "yes." We can get a representative sample using land-lines, but with response rates so low these days, it takes much longer than it used to. It is important to note that early studies involving cell-phone-only voters suggest their attitudes are not different from land-line voters in the same demographics, so missing them would not adversely affect the survey results. Further, incorporating cell phones into our samples does not really enhance the notion of "probability" sampling, since more cell users than land-line users opt out of participating in the survey.

Much like any common sales solicitation, the practice of conducting a poll by contacting a person on their cell phone used to be illegal. While those prohibitions have been swept away, many people still do not know that and are upset to receive an unexpected call from a stranger on their mobile phone.

There are ramifications to making such unexpected intrusions. Because they were not expecting the call, it is possible that they are not in a convenient place to take a call, so they may not feel free to give an honest and heartfelt answer to a pollster's questions. For instance, if you call someone at a crowded bus stop, they may not answer candidly. Or they may find themselves in loud surroundings and be unable to concentrate properly.

Add to that the invasive nature of a typical polling interview, which can involve 40 to 50 questions, and you begin to see how participating in a survey in a public setting could be very uncomfortable.

Then there is the matter of the respondent's frame of mind. Because of the way most cell phone calling plans are structured, the recipient of an unsolicited pollster call is likely to be put in a bad mood, which may very well affect their responses. They may also answer quickly without giving much thought to their responses, just to end the call earlier.

Yet another problem has to do with geography. While all phones have a geographic identifier known as an area code, cell phones are obviously much less limited to a certain geographic area. And now, because of legislation requiring the portability of cell phone numbers, a cell phone user could be anywhere, regardless of the area code associated with their phone.

All these problems aside, there are financial ramifications because the rate of completed surveys conducted with cell phone users is abysmal. Many more calls will have to be made to complete survey projects.

Any one of these problems could probably be overcome by a skilled public opinion research company. But when you consider the full range of complicating factors, any of which could adversely affect the accuracy of a survey (likely without detection until it is too late), the risk of polling cell phone users simply introduces too great an unquantifiable margin of error that could undermine public confidence in our research work.

Especially when you consider that the last two presidential elections have been won by razor-thin margins, this is far too great a risk.

John Zogby is president and CEO of the polling firm Zogby International. Comment about political topics on the forums at Zogby.com.