Response to My Gallup Query
"Presently only about 3 percent of the population uses a cell phone exclusively. In fact, this group does tend to be younger. We compensate for any differences this produces (as well as other omissions like those that might occur due to things like caller ID) by weighting our samples to national population totals for race, age, education, region, and gender. In fact, it may not be surprising that we get better participation in our polls from people with higher levels of education who in turn are more likely to embrace technology. A number of studies have been conducted about the impact of nonresponse, and they tend to find minimal differences, between surveys with very high response rates and those with lower ones. One of the more famous ones was published in Public Opinion Quarterly in the Summer of 2000. Another way to measure the accuracy of polls is to look at how well they match up to election totals (though election polls face additional challenges, such as how to consider undecided voters). In the 2002 congressional elections, Gallup's national estimates came within 1% each of the actual votes for Republicans, Democrats, and "Others".