Reply #32: The 2002 data could provide another test of your hypothesis, one that

[AirAmFan]

many statisticians might find more convincing. But we'd need to know what percent of voters in each state used e-voting with and without a paper trail in 2002 as well as in 2004.

Compare the (exit poll - actual results) discrepancy across time for every state and DC. Were discrepancies systematically bigger in 2004 in states that adopted e-voting with no paper trail between 2002 and 2004? That question uses each state's own past history of exit poll bias as a basis for comparison with the 2004 discrepancy.