paper on Florida

[dougrhess]

Anybody read this paper?

http://ucdata.berkeley.edu/new_web/VOTE2004/election04_WP.pdf

The data is at http://ucdata.berkeley.edu/

Looks like they used some data from this website. However, their OH numbers seem off for Cuyahoga county at least.

[jimrtex]

A problem is that 15 e-vote counties are concentrated in certain areas of the state, especially the most populous (including the 5 with highest turnout).  Florida has a large difference in county populations (the mean of the county Bush percentages in 2000 was 56.3%).

The counties with e-voting:

SE Florida (6): Indian River, St Lucie, Martin, Palm Beach, Broward, Miami-Dade.

SW Gulf Coast (3) Charlotte, Lee, and Collier.

Tampa-St Petersburg (3): Pinellas, Hillsborough. and Pasco.

Central (2): Sumter and Lake.

Suburban Jacksonville (1): Nassau.

If you look at areas where there is a mixture of systems, there doesn't appear to be that much difference.  For example, the shift in e-vote Pasco and not e-vote Hernando is almost identical.  Along the SW Gulf Coast, Charlotte and Manatee didn't use e-voting, and there is no clear pattern in their shift vs. the other 3.

It doesn't make sense that the _increase_ in the Bush percentage would be proportional to the 2000 Bush percentage.

While it does appear that many strong Bush counties did have large Bush increases, this may be due to other factors.   Almost all of these counties were in the more rural northern part of the State.  In other words, the voters were behaving like those in Alabama and Georgia.

Where does a shift come from?

(1) Converts.  But if a county is 70% Bush, then only 30% of the population is available to switch.  If you can convert 10%, you are up to 73%.  If a county is 30% Bush, and you convert 10% of the Gore vote, you are to 37%.

(2) Non-voters in 2000.  Is it reasonable to expect that these voters would be more pro-Bush than those who voted?  If anything, it might be reasonable to expect that they would be more neutral, but susceptible to conversion.  One reason for not voting is indifference to the candidates.  If you don't care who wins, why vote?  In other cases, a political minority in a locality may have become discouraged from voting.  If you can't influence the choice of legislators and county officers, you might be less inclined to turn out for presidential elections.  But after 2000, you would realize that you could have an influence on the presidential outcome.   So the new voters in strong Gore areas may tend to be more Bush-inclined than the repeat voters.  While those in strong Bush areas might be more Gore-inclined.

(3) Move-ins and move-outs.  These voters may not vote like the people who used to live in the area.  There are counties in Florida with 40% increases in turnout over 4 years.  Some of this is due to non-voters in 2000 and  other is from move-ins.

For example, Cubans moving north from Miami, may shift the balance of Broward county from 31.5% Bush to 35% Bush.  On the other hand persons moving to university counties like Alachua (U of Florida) and Leon (FSU & Florida A&M) may tend to hold the Bush vote down.  Service workers moving into the Orlando area may tend to vote more Democratic, especially if more established voters are moving into neighboring Seminole and Osceola counties.

The demographic factors that the study used in Florida may be meaningless with respect to politics.

There is almost no-correlation between personal income and the Bush percentage in 2000 or 2004 (0.063 and -0.058).   If you can't use income to predict the results of an income, is it likely that you can use it to predict the change in results.

There is a very weak correlation between between the Hispanic percentage and the Bush percentage in 2000 or 2004 (-0.222 and -0.263).  Hispanics in different parts of Florida have different origins.  In the agricultural areas west of Lake Okeechobee, many are migrant farm workers.  In the Miami area, many are Cuban, though many are from other areas in Latin America.  But the Cubans may be moving furhter north.

The authors ignored other demographic factors such as the Black population (Florida is in still somewhat part of the South).  The largest Bush decrease occured in Gadsden County, the only county in the state with a majority Black population.

There is a significant Jewish population in SE Florida.  I suspect many would identify more with Lieberman than Kerry, who emphasized his Roman Catholicism during the campaign.