The Impact of Cell Phone Non-Coverage Bias on
Polling in the 2004 Presidential Election
Scott Keeter
Pew Research Center
August 1, 2005
Abstract
Despite concerns that the accuracy of pre-election telephone polls would be harmed by the
omission of voters who could be reached only by cell phone, most national polls performed well
in predicting President George W. Bush’s reelection, and state polls were generally accurate as
well. The national exit poll conducted by the National Election Pool found that 7% of election-
day voters had cell phone service but no land line; younger voters were far more likely to be cell-
only: 19% among those 18-24 and 20% among those 25-29. Within these two youngest age
cohorts, cell-only voters were significantly more likely to be single and childless. While cell-
only voters were more supportive of John Kerry than voters overall, they were similar to voters
within their own age cohort. Because of this, pre-election telephone surveys that weighted their
data appropriately by age were not significantly biased by the absence of the cell-only voters.
1
Among the many challenges faced by polling organizations in the 2004 elections, the
newest and perhaps most frequently discussed was the growing number of people who relied
solely on a cell phone for telephone service. Cell-phone-only (CPO) households could not be
reached by most pollsters because major survey organizations do not include cell phone numbers
in their telephone sampling frames for political polls. Given the fact that CPO households
disproportionately include younger citizens, considerable speculation arose that their omission
would create a bias in pre-election polls. Many believed that the polls might understate John
Kerry’s support, given his greater popularity among younger voters.1 The potential danger of
CPO households for the polls was discussed in at least 150 separate news stories in major
newspapers, news magazines, or the broadcast media during the four week