|
OK, so there was a mean bias in previous exit polls that overestimated the Dem vote count (or elections have been rigged by the Repubs for years).
But again we are only dealing with simple averages.
To see if 2004 was really different than previous polls, we need to know more.
For example, what were the Median WPEs and the Absolute Mean WPEs in the previous polls? We have these numbers for 2004, from Mitofsky, and they show huge WPEs in both directions, with a small bias toward Kerry in the polls and/or toward Bush in the vote count.
What can we say about previous exit polls? Were the Absolute Mean WPEs just as high, well over 15%, and probably up to 30%?
Were there Median WPEs just as large, independent of precinct partisanship?
Without this sort of data, we can't make the assertion that the results of the 2004 poll were similar to or different from previous polls, except in its overall discrepancy from the vote count before adjustment. So, anybody got it?
|