This is the fourth in a series of state by state voter registration diaries that I have decided to do rather than the omnibus ones I had done in months past (the states I did diaries for before this one were for the states of North Carolina, Nevada, and Iowa).
By devoting a diary to a single state at a time, I hope that I can not only make the reading of the raw numbers more simple but more importantly, that the information is more in depth and more informative by being able to provide some commentary regarding those numbers for that state. In that vein I will try to provide more of a long term history of the voter registration numbers for each of the states, both over the past year and just as importantly in comparison to the final numbers from the 2004 election with accompanying raw vote totals from the election results in those states that year. In the case of Florida, I have decided to give an even longer historical perspective by looking at numbers going as far back as the 1996 general election (it should be remembered that Clinton won Florida in 1996).
That having been said onto the numbers for today's state Florida.
Dems Reps Ind
1/08 4,199,172 3,873,019 2,252,756
4/08 4,268,874 3,879,747 2,246,583
5/08 4,305,680 3,889,705 2,259,491
6/08 4,348,026 3,904,032 2,280,826
7/08 4,395,622 3,927,413 2,309,260
8/08 4,453,008 3,954,884 2,336,198
10/08 4,722,076 4,064,301 2,461,257
So over the course of the past ten months, Democrats have outregistered Republicans by adding roughly 522,904 new Democratic voters to the rolls to the Republicans adding roughly just 191,282 new Republican voters to the rolls, a factor of 2.7 to 1! Some interesting facts, in January, 2008, registered Democrats comprised 40.6% of the registered electorate in Florida and Republicans comprised 37.5% of the registered electorate, for a difference of +3.1% for the Dems (by the way Indies made up 21.8%). Now as of October, 2008, (after the registration deadline has passed and final numbers have been tallied) Democrats comprise 41.98% of the registered electorate while Republicans share has inched down to 36.13%, for a difference now of +5.85% for the Dems (Indies share of the electorate has remained flat staying at 21.8% of the electorate, a fact which I will highlight as a good thing later)!
Sounds great doesn't it, but before you get too excited it must be pointed out that Democrats have (at least for the past 3 elections going back to 1996) always outnumbered Republicans in the state and yet Republicans have outperformed Democrats come election day in the last two elections, either by performing at their percentage share of the electorate while Democrats performed below theirs (2000) or by actually overperforming Democrats outright (2004). This has important implications and (as explained below) may tend to diminish any usefulness that can be gleaned from looking to voter registration numbers, and the otherwise noticeable increase in terms both of absolute raw numbers and percentage share of the electorate Democrats have been able to show this year. But before explaining why this may (and I underscore may, as well be explained at the end) be, it is first important to look to the evidence suggesting this thesis.
The place to start being voter registration numbers from the past three elections in Florida.
So where was Florida in 2004? In short, registered Democrats outnumbered registered Republicans in the state. At the beginning of 2004, there were 3,903,563 registered Democrats, 3,568,697 registered Republicans, and 1,825,280 Indies; or for a percentage breakdown, Democrats accounted for 41.9% of the electorate, Republicans counted for 38.5%, and Indies made up 19.5%, not much different than where Florida is right now. By election day in 2004, there were 4,315,222 registered Democrats, 3,942,040 registered Republicans, and 2,189,546 Indies; or for a percentage breakdown, Democrats accounted for 41.3% of the electorate, Republicans 37.7%, and Indies 20.9%. Again except for the surge in Indie registration the two parties were not that far off from where they started at the beginning of the year (the Indie surge story will become important by the end of the diary). Yet despite this built-in advantage come election day in 2004, exit polls showed that Democrats only accounted for 37% of those who voted, Republicans made up 41%, and Indies made up 23%. In other words, Republicans outperformed, Indies outperformed, and Democrats significantly (from a statistical perspective) underperformed from their original share of the registered electorate. Interestingly, that year Bush won the nationwide popular vote by roughly 3% (a corrollation that, as explained below, had repeated itself in the two prior elections).
So where was Florida in 2000? Again Democrats outnumbered Republicans in terms of their share of the registered electorate. At the beginning of the year, there were 3,672,332 registered Democrats, 3,279,838 registered Republicans, and 1,338,825 Indies; or as a percentage, Democrats accounted for 44.2% of the registered electorate, Republicans made up 39.5%, and Indies 16.1%. By election day in 2000, there were 3,850,527 registered Democrats, 3,469,922 registered Republicans, and 1,549,698 Indies; or as a percentage of the electorate, Democrats accounted for 43.4%, Republicans made up 39.1%, and Indies made up 17.4%. Again much like 2004, there was an uptick of Indies registering with the two major parties share of the electorate decreasing slightly. Here, however, is the kicker, exit polls for 2000 (and yes everyone knows how flawed they were but it is all I have) showed that of the voters showing up to vote on election day 40% were Democrats, 38% were Republicans, and 22% were Indies. Again Democrats underperformed, Republicans basically treaded water, and Indies surged as a share of the voting public. Also again compare this difference in terms of share of voters coming to the polls, to the nationwide popular vote total between Bush and Gore, where Gore beat Bush by a little less than 1%. Hmm, could there be a corrollation developing between the margin in the share of the electorate showing up to vote in Florida between the parties and the margin in the nationwide popular vote? Well lets look to 1996.
So where was Florida in 1996? Just as before (or perhaps better said, just as would be the case with the next two elections) Democrats outnumbered Republicans in terms of the parties' share of the electorate, but by slightly bigger margins. At the beginning of 1996 there were 3,512,309 registered Democrats, 3,073,367 registered Republicans, and 847,629 Indies; or as a percentage share of the registered electorate, Democrats accounted for 47.2%, Republicans made up 41.3%, and Indies counted for 11.4%. By election day in 1996, the same pattern repeats itself, Indies increase their percentage share while Democrats and Republicans share decreases. By November 1996, there were 3,761,057 registered Democrats, 3,333,407 registered Republicans, and 1,065,081 Indies; or as a percentage share of the electorate Democrats made up 46%, Republicans accounted for 40.8%, and Indies counted toward 13%. The exit polls for election day 1996 in Florida showed Democrats making up 45% of the voters who voted that day, Republicans made up 34%, and Indies 19%. Again, Indies outperform their share of the electorate, Democrats tread water, and Republicans woefully underperform (almost the inverse of the 2004 election but with bigger margins). Interestingly, the nine percent difference in terms of share of voters showing up to vote in Florida was nearly matched by Clinton's nationwide popular vote spread with Dole with Clinton beating Dole 49% to 41%.
It is this eerie connection between share of the voting electorate on election day and spread in the nationwide popular vote that I have not seen in other states. The only corrolation between voter registration gains throughout the year and share of the voting electorate appears to be limited only to Indie voters in Florida. For some reason, their numbers seem to increase both in raw numbers and in terms of percentage of registered elecorate in each of the past three Presidential elections in Florida, while the two major parties share in terms of percentage (their raw numbers do increase) tends to be one decreases and the other generally treads water there being no rhyme or reason for which party goes down and the other stays the same.
This connection could suggest that voter registration trends simply will not provide a good indicator of which side's voters are going to show up and in what percentage. Rather, under this thesis the best measure for calculating party performance (in terms of percentage share of voters showing up on election day) in Florida is to look at the national polls that do not weight by partisan id (i.e., do not look at Rasmussen) and then average out those polls partisan id numbers and apply that breakdown to Florida. If Democrats have the advantage in the nationwide popular vote, then that is good indicator that they will also enjoy a similar advantage in Florida in terms of the spread in the percentage of the voters who show up, and vice versa.
There is something, however, that is keeping me from buying into this thesis for this year's election. Look at the Indies registration numbers in FLA this year. Indies percentage share of the electorate has stayed flat this year, it has not increased as it has done in all three prior Presidential elections. Republicans percentage share on the other hand decreased as it has in all three past Presidential elections, not by much but a drop off nonetheless. Now look at Democrats percentage share. It has actually INCREASED this year! In none of the other three past Presidential elections has a party's share (Republican or Democrat) of the electorate gone up before election day. That this has occurred may suggest a Democratic surge will occur in Florida, much along the lines of Indies share of those who voted has upticked in elections past by anywhere from 4 to 6% over their actual percentage of registered voters! Could it be that Democrats are acting like the Indies from Presidential elections past in Florida this year? If so, this would be a huge lift to Obama's chances of winning the state, and an equally large blow to McCain's chances. Given that this trend has continued throughout the past year gives me some comfort in believing that Obama will have a good night in FLA come election day.