The Unexplained Poll Discrepancy

Steven F. Freeman
stfreema@sas.upenn.edu
December 29, 2004

Most Americans who listened to radio or surfed the internet on election day this year sat down to watch the evening television coverage thinking John Kerry won the election. Exit polls showed him ahead in nearly every battleground state, in many cases by sizable margins. Although preelection day polls indicated the race dead even or Bush slightly ahead, two factors seemed to explain Kerry’s edge: turnout was very high, good news for Democrats, and, as in every USpresidential election with an incumbent over the past quarter-century, undecided voters broke heavily toward the challenger.

But then, in key state after key state, counts showed very different numbers than the polls predicted; and the differentials were all in the same direction. The first shaded column in Table 1.1 shows the differential between the major candidates’ predicted (exit poll) percentages of the vote; the next shaded column shows the differential between their tallied percentages of the vote. The final shaded column reveals the “shift.” In ten of the eleven consensus battleground states,the tallied margin differs from the predicted margin, and in every one, the shift favors Bush.

The media have largely ignored this discrepancy (although the blogosphere has been abuzz), suggesting either that the polls were flawed, or that the differential was within normal sampling error, a statistical anomaly, or could otherwise be easily explained away. In this report, I examine the validity of exit poll data, sampling error, the likelihood of statistical anomaly, and other explanations thus far offered to explain this discrepancy.