Here’s the racial/ethnic breakdown by party for the first day of early in-person voting in Florida.
Of the 8,377 Democrats who have cast EIP, 52% are white, 31% are black, and 13% are Hispanic. Some 1,666 Democrats in Hillsborough County have cast EIP ballots, 57% of them white, 29% of them black, and 10% Hispanic. In Miami-Dade, whites have cast 37% of the EIP Democratic ballots, blacks 34%, and Hispanics 26%. In Duval County, with 1,439 EIP Democratic ballots cast, blacks have cast 49% and whites 45%. And finally, in Orange County, 47% of the EIP Democratic ballots cast thus far are white voters, 28% are black, and 18% are Hispanic.
Among Republicans, of the 9,354 Republicans who have cast EIP, 81% are white, 1% are black, and 15% are Hispanic. In Miami-Dade, where Republicans have cast 1,543 EIP ballots, 68% have been cast by Hispanics, and only 28% by whites. In Hillsborough County, where the most EIP ballots have been cast by Republicans (1,757), 89% of the ballots cast thus far are by whites. White Republicans casting EIP votes in Duval make up an even greater share of the 1,511 votes cast, some 94% of the total. In Orange County, 10% of the EIP votes by Republicans were by Hispanic voters, with white voters comprising 85% of the total (1,385) EIP votes cast on the first day of early voting.
There was a slight uptick in the number of active registered voters in Florida immediately prior to the registration deadline to be eligible for the March 15 primary. From January 31 through February 16, 2016, the voter rolls grew by 62,318 voters. Certainly the number of newly registered voters was higher than 62k, given that the voter rolls are dynamic; Supervisors of Elections regularly remove voters from the rolls, including those who are deceased, move out of state, or are convicted of felonies. Under federal law (NVRA), SOEs should not be removing inactive voters during this period of time, given the immediacy of the March 15 presidential primary.
Is the increase in total registrations distributed evenly across racial and ethnic groups? No.
Hispanics now comprise 14.88% of the electorate, up .04% from the percentage of Hispanics in the January 31, 2016 active voter file. On the other hand, blacks now comprise 13.31% of the active voters, down from 13.34% of the January 31 active electorate. The percentage of the Florida electorate that is whites is also down, from 65.74% to 65.71% of the electorate.
These are pretty steep changes for just a 15 day window of new voter registrations, and likely reflects broader demographic changes in the state (more younger Hispanics eligible to register to vote) and general attrition rates from the voter file of white and black registered voters who have been removed from the statewide voter file.
What about these last-minute changes across the parties? Some interesting patterns emerge.
Registrants (with the possible exception of Jeb Bush) are generally not likely to change their racial or ethnic categorization on a voter registration form. But current registrants might very well might change their party registration ahead of a closed primary contest if they plan on voting. This is especially true of No Party Affiliates (NPAs), who are excluded from voting in party primaries in Florida.
This very well might explain the drop in total NPA registrations in Florida over the two week period prior to the February 16 registration cutoff, from 2.892m to 2.878m active NPA voters. The decline of 13.6k registered voters is not Huuuuge in the grand scheme of things, but it is significant, given the general trend in the Sunshine State of more voters registering as NPAs over the past two decades.
It bears noting that nearly all of the decline in NPA registered voters appears to be due to the decline in white NPAs; the total number of registered Hispanic NPAs, and even black NPAs, increased over the two-week period.
Although there is certainly the possibility of an ecological fallacy at play when interpreting these aggregate numbers as a sign of NPAs engaged in last-minute party-switching, there was an uptick in the number of Republican active registered voters, nearly 50k from January 31 through February 16, to 4.276m. Democrats also increased their rolls, but less by than 30k, to 4.570m for the Presidential Preference Primary book closing.
In light of several news articles that have reported the findings of a recent joint NCSL & PEW Report, “Who We Elect: The Demographics of State Legislatures,” I thought it might be useful to convey some points made by Dr. Carl Klarner regarding the inherent problem when comparing “average” state legislative racial and ethnic (and other socio-demographic data) with national averages. As Klarner writes in his recent Report (p6),
“state level figures [should] always [be] weighted by state population when being averaged to the nation level. This is done to approximate the importance of a particular state legislative chamber….
The issue of weighting by population is especially relevant to assessing minority representation, as it makes a big difference. Reports of minority state legislators frequently state them as a percentage of all legislators in the country.17 It is also sometimes asserted that descriptive representation is lower in state legislatures than in the U.S. House. But it is important to take the varying size of state legislative districts into account. For example, there are currently 498 African-American state house members, or 9.2% of the 5,411 state house members in the United States. But African-American state house members actually represent 11.9% of the United States, if you factor in the size of the districts they represent. This is substantially closer to the 13.7% of the United States that is African-American.18 The simple reason for this is that Northeastern states, with unusually large state houses, have few African-Americans living in them. Looking at minority descriptive representation in this way gives us more insight into what types of electoral arrangements are conducive to the fair representation of racial minorities.”
I agree with Dr. Klarner. Weighting state legislators by population is essential when comparing national averages with minority representation across the 99 state legislative chambers.
According to the 2010-2014 5-year averages of the American Community Survey (ACS), conducted by the U.S. Census, African-Americans make up approximately 13.7% of the national population, Hispanics 16.9%, and Asians 5.9%.
According to Dr. Klarner, here are the percentages of current state legislators who are racial/ethnic minorities, unweighted v. weighted:
State House Members Unweighted Weighted
- African-American: 9.2% 11.9%
- Latino 4.0% 8.5%
- Asians 1.5% 2.3%
State Senators Unweighted Weighted
- African-American: 8.3% 10.1%
- Latino 3.6% 6.9%
- Asians 1.7% 1.9%
Here is the abstract:
Shortly after the Supreme Court in Shelby County v. Holder struck down Section 4(b) of the Voting Rights Act (VRA), the state of North Carolina enacted an omnibus piece of election-reform legislation known as the Voter Information Verification Act (VIVA). Prior to Shelby portions of North Carolina were covered jurisdictions per the VRA’s Sections 4 and 5—meaning that they had to seek federal preclearance for changes to their election procedures—and this motivates our assessment of whether VIVA’s many alterations to North Carolina’s election procedures are race-neutral. We show that in presidential elections in North Carolina black early voters have cast their ballots disproportionately in the first week of early voting, which was eliminated by VIVA; that blacks disproportionately have registered to vote during early voting and in the immediate run-up to Election Day, something VIVA now prohibits; that registered voters in the state who lack two VIVA-acceptable forms of voter identification, driver’s licenses and non-operator identification cards, are disproportionately black; that VIVA’s identification dispensation for voters at least 70 years disproportionately benefits white registered voters; and, that preregistered sixteen and seventeen year old voters in North Carolina, a category of registrants that VIVA prohibits, are disproportionately black. These results illustrate how VIVA will have a disparate effect on black voters in North Carolina.
Race, Party, and the Consequences of Restricting Early Voting in Florida in the 2012 General Election
Political Research Quarterly
Michael C. Herron and Daniel A. Smith
In mid-2011, the Florida legislature reduced the state’s early voting period from fourteen days to eight and eliminated the final Sunday of early voting. We compare observed voting patterns in 2012 with those in the 2008 General Election and find that racial/ethnic minorities, registered Democrats, and those without party affiliation had significant early voting participation drops and that voters who cast ballots on the final Sunday in 2008 were disproportionately unlikely to cast a valid ballot in 2012. Florida’s decision to truncate early voting may have diminished participation rates of those already least likely to vote.
OnlineFirst: February 24, 2014
Available: Full Text (PDF)
My colleague, Michael Herron at Dartmouth, and I have just finished crunching the 2012 General Election statewide voter file.
We’ll have lots to report in the coming days about the racial and ethnic voter participation in the November election, including statewide and county breakdowns for early voting and absentee voting. We’ll also have some data to report on the rejection rates of provisional ballots and absentee ballots across racial and ethnic groups.
But for now, one item that caught my eye this morning was the considerable inflation of supposed Latino voter participation in the 2012 General Election.
13% African American
Further analysis of the Florida exit polls conducted by the Pew Research Hispanic Center immediately after the election, “Latino Voters in the 2012 Election,” found that “Hispanics made up a growing share of voters in three of the key battleground states in yesterday’s election.” According to the report, “Hispanics made up 17% of the Florida electorate this year, up from 14% in 2008.” The Pew Report continued:
The state’s growing non-Cuban population—especially growth in the Puerto Rican population in central Florida—contributed to the president’s improved showing among Hispanic voters. This year, according to the Florida exit poll, 34% of Hispanic voters were Cuban while 57% were non-Cuban. Among Cuban voters, the vote was split—49% supported Obama while 47% supported Romney. Among the state’s non-Cuban voters, Obama won 66% versus 34% for Romney.
Yet, when matched against the Florida Division of Election’s December 31, 2012 voter file, our analysis suggests that the 2012 exit poll estimates considerably over-inflate the actual Latino makeup Florida’s 2012 electorate.
In 2012, roughly 8.43 million Floridians cast ballots in the General Election.
According to our analysis of the state’s voter history file, a little more than 1 million citizens who self-identified on their voter registration cards as Latino voted in the 2012 election. That’s only 12.5% of Florida’s 2012 electorate.
In contrast, nearly 14% of Florida’s 2012 actual electorate was African American, close to a full percentage point greater than the exit poll estimates. White voters were similarly under-represented in the exit poll estimates, as slightly more than 68% of Florida’s 2012 electorate was white. (Incidentally, Floridians who voted in the 2012 General Election and who identified as “Other” or “Multi-racial” on their voter registration cards tallied less than 2% of the vote in 2012.)
For now, I will leave it for others to interrogate why the 2012 Exit Polls considerably over-inflated Latino turnout in Florida, but I have some suspicions that I will offer down the road as time permits.