Steve Freeman: Polling Bias or Electoral Fraud? An Examination of State-Level Discrepancies Between the Official Count and Exit Poll Results in the 2004 US Presidential Election"
Abstract: Exit poll data, based on 114,559 sampled voters at 1,460 precincts across the nation who were asked in confidential questionnaires how they voted, suggest that John Kerry won the 2004 US Presidential election. These survey data deviate 6.5 percentage points from the official counts in those same precincts (Within Precinct Error or WPE), a discrepancy which extrapolates to eight million votes nationwide.
The official explanation for the discrepancy – that Kerry voters participated in the exit polls at a higher rate than Bush voters – is unsubstantiated by evidence. The very data presented to justify the “reluctant Bush respondent” thesis seems to contradict it entirely. Unfortunately, the National Election Pool has refused access to the precinct level data that could be used to properly investigate claims of polling bias and electoral fraud.
In this presentation, I use publicly available, state-level data to investigate theories put forward to explain the discrepancy. Statistical analyses indicate no support for the “reluctant Bush respondent” thesis; rather, the data suggest that the exit polls slightly overstate support for the President. Data do reveal, however, statistically significant correlations between WPE and Republican gubernatorial control, state electoral importance, voting technology, and other election administration variables. These relationships are inconsistent with theses of polling bias, but consistent with theses of electoral fraud.
Warren J. Mitofsky: “2004 Exit Polls: What Bloggers And Others Got Wrong”
Abstract: The 2004 exit polls were generally believed to have indicated a victory for John Kerry. Nothing could have been further from the truth. This was compounded by the conspiracy theorists after the election who mistakenly claimed the exit polls validated their claim. There was no evidence in the exit polls to substantiate these claims. On election day the misinformation about the exit polls was spread by inexperienced people trying to make sense of complex statistical data. After the election the academics who yelled election fraud were no better. It is time to set the record straight.