Reply #7: This time I think you got it wrong. Its not just about statistics, but as [View All]

[berniew1]

assumptions and history also. The touch screen counties have very different characteristics and voting histories than the optical scan counties. The touch screen counties are all big coastal mostly urban counties(with sprawl) and a lot of tourism,etc. They have well documented switching and suppression problems in 2004 election. The optical scan counties are a few large urban or coastal counties like Duval(Jacksonville), Orange(Orlando), Brevard(Titusville/Melborne), Leon(Tallahasse), Alachua(Gainesville) and the rest are all small rural counties that have a historic conservative voting pattern in national elections. So unless you believe there has been something amiss in past elections as well as this one, the small rural county sector can't be dealt with using your statistical methods based on assumptions that Dems vote for Dems. Bay County(Panama City) and Escambia(Pensacola) are also optical scan but are in the panhandle and vote like the north florida rural counties and Alabama.

Tallahassee and Gainesville are dem majority areas with good electoral systems, Duval was the worst county from 2000 and pretty bad in 2004, Orange and Brevard appear to have also had significant problems and swings in 2004. But there is no basis I've seen for assuming a swing in the small rural(bible belt) counties.
I think your analysis is totally off base in this case. Since you don't have exit poll data for the small rural counties, the best info I know is the 2000 official vote data- which is given in my file. Are you suggesting there was a major problem with the official 2000 votes in those counties in 2000?? Its possible, but I haven't seen anyone make the case. Several of them have Dem SOEs.

See the analysis: http://www.flcv.com/fla04EAS.html
and the data for each county for 2000 and 2004
http://www.flcv.com/fla04EA.html

The big urban counties have the majority of the population, but I think you have to treat the urban counties and the rural counties separately, and use different assumptions for them based on EP or history.