The fight for marriage equality had a dramatic shift this summer. The Supreme Court decision, United States v. Winslow, struck down Section 3 of the Defense of Marriage Act; meaning state with same-sex marriage offered the couples with those licenses access to federal benefits. However, with no Supreme Court decision legalizing same-sex marriage nationwide, pro-equality groups are left looking to other state opportunities. Efforts are in full gear for a referendum in Oregon, and Illinois will be targeted for legislative approval in the 2014 session. One state that has been subject to debate is Florida. The swing state that gave Barack Obama its electoral votes twice has factors that make it a strong candidate for support; with other factors that make the chances of success unlikely in the short term. The state’s growing Hispanic population, a group that was opposed to same-sex marriage in 2008, but has shown a dramatic shift toward support; coupled with liberal white populations in Leon, Alachua, Miami-Dade, and Broward counties, give the impression the state is prepared to endorse marriage quality. However, the two biggest obstacles toward marriage equality in Florida are its large retiree population as well as the 60% threshold the pass a constitutional amendment. The 60% threshold is undoubtedly the biggest obstacle for marriage equality in the sunshine state.
Same-sex marriage has already found itself on the Florida ballot before. In 2008 the state voted to send Obama to Washington, but also voted to ban same-sex marriage with 62% of the vote.
Only one county, Monroe, rejected the ban. The support for the ban in Leon, Broward, and Alachua counties was the cause for the most concern among equality activists. In the fast-paced shifts in support for marriage equality, 2008 was still a period where a majority of Americans were against the practice, However, the support for the ban in some of Florida’s most liberal counties showed just how much persuasion needed to be done.
The reason for these counties supporting the bans fell to one over-arching issue; minority populations support the ban. In all the mentioned counties, Hispanics and African-Americans supported the ban on cultural and religious grounds. In a state with large minority populations, this spelled doom for marriage equality in 2008.
Lets look at this 2008 vote on a more localized level. Doing so will allow us to see voting patterns in a clearer and more refined manor. It will also allow us to get a glimpse at the future of same-sex marriage in the state when the issue comes to a vote again. For this local look, I have chosen my home county of Broward. The county is the second-largest in the state and the most reliably Democratic urban county; often giving Democratic candidates well over 65% of the vote.
Broward County in 2008: Race and Partisanship
In 2008, Broward County gave Obama 67% of the vote, sent several Democrats to congress and the state legislature, and maintained an entirely Democratic county commission. The county’s demographics of white suburban liberals, Hispanics, and blacks, have made the county the most Democratic county in the state; with the exception of majority African-American Gadsden.
NOTE: I will try to constantly use the phrase “black or African-American” in this article to account for the fact that a large percent of the black population of Broward are immigrants from the Caribbean and thus do not consider themselves African-American. Any reference to “black”, “African-American”, “black or African-American” is meant to include the entire population that would count themselves as black/African-american during the 2010 census. Shifting in the phrase used is done for cosmetic purposes.
The county has had a constant influx of Hispanic and Black immigrants over the last several decades. Immigrants from the Caribbean Islands and Central America are especially prevalent in the southern ridge of the county. Spanish and Haitian Creole are sizable minority languages in the county. As of the 2010 Census, the white population was below 50%.
Hispanic voters have seen the largest growth and in all likelihood will soon surpass black or African-Americans in the county. The figures above are the racial makeup of the Voting Age Population of the county.
While white voters fall well below 50% overall, the registration numbers tell a different story. Registration numbers around the same time as the census showed Hispanic registration much lower (as a percent) than the overall population.
Hispanic drop-off was 9% from its total VAP population. This was most prevalent in the city of Weston (on the central western edge of the county). Part of this drop-off can be attributed to a significant number of VAP minorities being immigrant workers with green cards but not yet citizens. However, in cases like the suburban city of Weston, the phenomenon can be tied to the nationwide issue of Hispanic voters being less involved in the political process and not even registering.
Democrats maintain such an advantage in the county because in addition to sizable minority populations, the white voters of the county are overall Democratic leaning and more liberal than white voters statewide. During the 2008 election, white voters in Broward backed Barack Obama. This trend held in 2012, but I am focusing on 2008 since this is when the same-sex marriage ban was voted on.
Lets start off with a look at the 2008 Presidential Election in Broward
Obama won an overwhelming number of the precincts
The map below shows the white precincts of Broward and who they voted for in 2008. Every non-white precinct cast a ballots for Obama.
Suburban and working class whites backed the President in 2008. Older white voters who live along the beaches generally cast ballots for McCain; as did older whites in Davie and the rancher-style town of Southwest Ranches on the counties western side. Especially wealthy whites voted for McCain in some places and Obama in others.
All this combines to make Broward a Democratic stronghold. Such a county would have been a strong candidate for rejecting the ban of same-sex marriage. However, when the vote took place, the county backed the ban with 52% of the vote. The breakdown is below.
Support for the ban is colorized in shades of red, while rejection is colored in green.
Consulting the precinct-level racial map shows that the non-white precincts are showing support for the ban in the above image. Looking at the racial correlation with a vote for or against the ban tells us a lot about Broward’s support for the ban.
Before going further into the precinct-level results, lets look at the results by the cities themselves. Broward is almost entirely Incorporated into 31 cities or towns, with very little land outside an organized municipality.
The Wikipedia page for Broward shows a map of the cities and the name for each. http://en.wikipedia.org/wiki/Broward_County,_Florida
With the exception of Weston, which is estimated 1% more VAP Hispanic than White, all the cities that rejected the ban were majority or plurality white. Cities that supported the ban hardest were either conservative white (Southwest Ranches or Sea Ranch Lakes) or black/African-American majority (West Park, Miramar, Lauderdale Lakes, North Lauderdale). Hispanics, concentrated in west Miramar, Pembroke Pines, and western Weston, all backed the ban as well. Hispanics make up a smaller segment of the voter rolls than blacks or African-Americans, so their impact on the ban is not as strong as African-Americans’. However, Hispanic’s support for the ban is clear. One good example is to look at the precincts that make up Weston. Weston narrowly rejected the ban, with its whitest precincts being the primary source of opposition. The precincts that were less white, and had higher concentrations of Hispanics, supported the ban.
Meanwhile, white suburbs like Cooper City, Coral Springs, and Tamarac rejected the ban; along with urban cities like Ft. Lauderdale and Hollywood. Wilton Park and Oakland Manors, known to be two of the largest gay per-capital cities in the nation, overwhelmingly rejected the ban. Several white suburbs like Davie and Hollywood narrowly supported the ban thanks to their older population.
Lets keep delving into racial voting patterns with the ban. The map below shows how precincts that were majority or plurality a specific racial group voted for the same-sex marriage ban; using the same red/green color scheme to indicate support or rejection of the ban.
A vast majority of precincts that rejected the ban were white, with only a small handful being black or Hispanic. In general white precincts split on the ban. A lot of information can be derived from these racial breakdowns; lets start with some basic facts.
- The Pro-Ban precincts are 45% white, 28.3% black, and 17.7% Hispanic
- The Anti-Ban precincts are 71% white, 8% black, and 12.5% Hispanic
Precincts that rejected the ban were overall much more white. These precincts gave Obama 63% of the vote, while the precincts that supported the ban actually gave Obama 68% of the vote. Blacks and Hispanics, two groups that overwhelmingly backed the President, also gave the ban high margins of support.
- The mean vote in favor of the ban for white precincts was 48%
- The mean vote in favor of the ban for Hispanic precincts was 60%
- The mean vote in favor of the ban for black precincts was 61.7%
The following six images were further highlight the racial breakdown of the precincts when it comes to the ban on same-sex marriage. The first three are histograms, which show the how many precincts voted for a certain percent on the ban. The blue lines indicate the 50% line. Notice how white precincts split. The numbers on the left side indicate the raw number of precincts that fall under the percentage line on the x-axis.
The histograms provide a clear visual of how the different racial precincts voted. Black precincts largely voted in the high 50s and mid 60s range for the ban, while white precincts were largely in the swing region of 45 to 55 percent.
Histograms can easily visualize how precincts broke down in the vote. Scatter-plots, meanwhile, can show how precincts voted in correlation with other factors. I graphed the vote in favor of the ban with the three racial characteristics of the precincts. I will go through each result one at a time.
First, how precincts voted based on how white a precinct was.
The plot shows the percent for the ban on the left hand side, and the percent the precinct was white on the bottom axis. The blue lines serve as dividers between 50% white and 50% for the ban. The scatter-plot indicates that as precincts became more white, they shifted toward rejecting the ban. The statistical correlation is not overwhelming (as the line is not incredibly steep, and other statically models, including correlation of coefficient, indicate a weak, but noticeable, trend). The trick is that overwhelmingly white conservative precincts supported the ban, weakening the position of the white precincts that in some cases overwhelmingly rejected the ban (the bottom, right box). Still, there is a clear trend that as precincts became white they reduced their support for the ban.
Next lets look at precincts based on how black or African-American they were.
The scatter-plot shows the opposite trend than white voters, showing that as a precinct becomes more black or African-American in general, its support for the ban grows. Only a handful of precincts that were majority black rejected the ban while a vast majority supported it. However, the correlation is still considered “weak,” statistically speaking. Notice that support for the ban plateaus in the 60s even as precincts become more black. While a precinct was more likely to support the ban as it became more black, fueled by the communities religious and cultural opposition to same-sex marriage, the opposition only went so far.
Finally, looking at how precincts voted based on how Hispanic precincts were
The scatter-plot of Hispanics doesn’t tell us much because no precinct falls above the 60% Hispanic line. Again, the general trend-line is a growth for the ban as precincts become more Hispanic.
When working with a county like Broward, which is home to three major racial groups, data must be analyzed carefully. For example, several precincts that were either plurality Hispanic or had a large Hispanic population, also had large black/African-American populations. Black voters had a clear support for the ban and Hispanic-heavy precincts could be argued to have voted for the ban because of the black voters. The scenario could also apply with Hispanics in the town of Southwest ranches. The precincts of that town supported the ban and voted for McCain; but that does not mean that was aided by Hispanics; instead that can be attributed to the conservative, wealthy, white population. The crucial data that indicates that indeed Hispanics supported the ban comes from polling history and from precincts in democratic Weston and Pembroke Pines; where whiter precincts rejected the ban but supported it as Hispanic numbers alone grew.
To highlight this, I selected the precincts of Weston and Pembroke Pines and subtracted any precinct that was more than 10% black or African-American. This left me with precincts that still had significant Hispanic registration and would be focused on two races rather than three. The scatter plot of the results is below.
The results show a strong statistical correlation between a precincts support for the ban and its Hispanic registration. The sample size is fairly small, 39 precincts, yet it gives a clear trend. This verifies that overall Hispanics also supported the ban in Broward.
Minority voters backed Obama yet they also supported the same-sex marriage ban. The result was that a large number of Obama-favored precincts voted for he ban, as can be seen below. Only a few precincts that backed McCain ended up voting against the ban.
Sixty percent of the precincts that voted for President Obama also support the ban.
What we see here is the same coalition that united to vote for the President in 2008, split on the issue of same-sex marriage; allowing the ban to succeed.
Education, Age, and Nationality
Race has revealed a great deal about why Broward voted the way it did. However, I took the liberty to look at other factors as well. Looking to the census’ demographic and soci-economic data, I looked for other possible factors in the vote.
The first test was education and income. The income of an area did not seem to have any significant correlation with the vote on the ban. All the economic groups had voters that supported or rejected the ban and there was no clear correlation as census tracts got more or less wealthy.
Education, as polling has verified, had a more significant correlation with a vote for or against the ban. Education did not provide a clear correlation, but the extremes of the education spectrum showed a clear distinction in support for the ban.
- Census tracts considered well educated, in that 50% of the population or more had a Bachelors Degree or higher, only gave the ban 45%.
- Meanwhile, tracts with less than 20% Bachelors Degrees or more gave the ban 58% of the vote; a 13 percent gap.
The first map shows the census tracts categorized by the percent the population had a Bachelors Degree or more. The second map shows the boundaries of the tracts that were the most or least college educated imposed over the precinct results for the vote on the ban.
Education’s influence on support for the ban matches polling that shows a more education populace will be more supportive of gay rights. Education can be one factor explaining why white voters supported or rejected the bans. Almost all the high education census tracts that fell over white precincts were ones that rejected the ban. However, the different boundaries that tracts and precincts employee makes direct comparison imperfect. In addition there is much less consensus in areas that are between 21% Bachelors and 49% Bachelors. Education appears to have played some factor, but could easily be overridden by others.
Age was another demographic characteristic I examined. Polling has shown one overarching trend-line when it comes to age; younger people support gay marriage and older voters do not. These trends continue to evolve, but were absolutely the case in 2008. There is not much to add on that subject. Broward has an elderly population, but not one as clustered as in areas like Palm Beach and the Gold Coast. The map below shows census tracts with the percent of its population that was 60 years or older.
Elderly population is especially clustered on the coast, which split on the ban, and a large amount of the tracts had populations less than 20% 60 years or older. Even some of the oldest tracts split on the ban. Overall older voters supported the ban, perhaps by a lesser margin in Broward. However, the exact correlation is not easily visible with the present data.
Finally, I examined a demographic characteristic similar to that of race itself. Broward and Miami-Dade are both home to large numbers of racial minorities that are immigrants or naturalized citizens. Immigration from South/Central America and the Caribbean Islands are major sources of minority growth in Broward. The map below shows tracts with the percent that is naturalized citizens.
Compare it to the map of the ban vote and it shows that these areas supported the ban. These same areas are often more than 50% Hispanic or Black, so this does not tell us anything new in that respect. However, it gives a further insight into why minority voters supported the ban. Religion, especially Catholicism, is a major force in countries Broward immigrants are coming from, and is further explanation for the support in the ban by minority voters.
Where is support/opposition Clustered?
We have seen several variations of the precinct-level maps of the vote on the ban. However, it can be hard to see where hot-beds of support and/or opposition are located. Up to 40% of the precincts in Broward voted between 45% to 55% for or against the ban. However, there are clear clusters of support and opposition within the county. Seeing where such clustering is taking place is important because it gives us the ability to see where there is strong community support or opposition. It is one thing for a single precinct to reject the ban by a heavy margin; but to see 10 precincts all next to each other do the same thing gives us the indication of a political community that is united on the issue. Knowing where these communities are will be important as the next campaign for same-sex marriage begins.
To best visualize these clustering, I used what’s known as a ‘hot-spot analysis’ tool in my GIS software. The simple summation of the tool is as follows: the tool records every precinct’s percentage in favor of the same-sex marriage ban. Next, it calculates how much each precinct’s percent deviates from the mean (average) of all the precincts combined. This deviation is measured in how many standard deviations away from the mean (52%) each precincts value is. The tool also factors in precincts being next to each other (as a method of examining the effects neighboring precincts have on each other’s voting patterns). The result will be a map with precincts colorized based on how they performed under these criteria. Precincts are categorized by how many standard deviates they are away from the mean (see the legend in the below map), and if their neighbors also deviated to a similar degree. The idea is to see where clusters of high deviation from the mean took place. The map below shows precincts colorized red to indicate clusters of deviations away from the mean and in favor of the ban (a positive number). The blue precincts show high deviation against the mean (a negative number), rejecting the ban. The white precincts only deviate 1.65 or -1.65 standard deviations away from the mean or are next to neighbors that do not deviate far, thus those precincts are considered to not be part of any mass-clustering of support or opposition. The software calculates the probability any clustering is statistically significant or if its just random chance; with those precincts considered random not colorized blue or red . A precinct that is 35% against the ban, next to one that is 65% for the ban, will be considered random and thus not be colorized; falling into the middle, random, category. Any precinct colorized red or blue is considered to have at least a 90% likelihood that the clustering is significant, that it is part of a trend of that grouping or precincts. The lightest color shading is 90%, the medium shade is 95%, and the darkest shade is 99% likelihood.
The result is below
The results tell us a good deal. The blue precincts are clusters of precincts that showed clear rejection of the ban; while red precincts supported and clustered in favor of the ban. The blue precincts all fall in largely white areas of the map; while red zones are in minority-heavy areas as well as conservative white precincts. The grey zones had too much randomness and little uniformity or strong deviation from the mean. What this map shows us is where communities of strong support of opposition for gay rights resided in 2008. One key value to this map? As signature gathering begins, focusing on the blue precincts will likely yield high positive response rates.
An important piece of info about the above map. In either clustering group there is likely a precinct or two that went against their neighbors and voted the opposite way on the ban. However, the software overrides those few precincts if they are being overwhelmed by their neighbors. A pro-ban precinct can still be colored blue is it is in the middle of a massive community of anti-ban precincts. The clustering is meant to show communities of support or opposition; but their is always some dissent within a community.
A few facts about these two clusterings of support and opposition
- Pro-Ban clusters were 40% white, 33.3% black, and 17.5% Hispanic. The precincts gave Obama 72.6% of the vote.
- Anti-Ban clusters were 68.5% white, 13.8% black, and 10.9% Hispanic. The precincts gave Obama 65.6% of the vote.
Race has been a clear determination of support or opposition to the ban. With that it mind I also created a hot-spot analysis of race in Broward. I decided to keep it simple, doing a hot-spot of just white voters. The red clusters indicate heavy white clustering, the blue indicate heavy non-white clustering; which will account for Hispanics and Blacks.
There is a good deal of racial clustering in Broward.
- These white clusters gave the ban 47.5% of the vote.
- The non-white clusters supported the ban with 57.3% of the vote.
A look to the future
Since 2008 there has been a growing support for same-sex marriage across the county; and Florida is no exception. The President’s support for the measure has increased Democratic and African-American support while doing nothing to decline the tepid but slowly growing support among Republicans. I decided to use polling data, combined with the 2008 vote in Broward County to see just how much support has increased and to show what a vote on same-sex marriage might look like if a vote happened today. Those of you who have followed my website may remember me doing a similar experiment in Leon County. You can see the article here, Support for Same-Sex Marriage Grows in Leon County.
First, I went through the old polls from Public Policy Polling, a reliable firm that proved startlingly accurate in the last several election cycles. Their polling showed trends toward same-sex marriage over the last few years, but I focused on two specific polls. I took a polling memo from July 2011 and another from June 2012 (after Obama’s same-sex marriage announcement. The trend-lines are as follows
- In 2011 support for gay marriage was 37% with 53% against. African-Americans rejected gay marriage 33% to 56% and white rejected it 37% to 53%. Hispanics also rejected same-sex marriage 31% to 55%.
- By 2012 support for gay marriage was 42% in favor with 45% against (with many moving to undecided). African-Americans supported gay marriage 49% to 31%, Hispanics supported 43% to 31%. and whites narrowly opposed 43% to 49%.
This shows a dramatic swing with African-Americans, who came to support gay marriage 16 points more than before. White support increased by 6 points and Hispanics by 12.
Now, these results are for a statewide poll, not Broward County. However, we can use those polling figures to get an estimation of how Broward County voters have evolved on same-sex marriage. Specifically looking at race, we see a strong increase in support for same-sex marriage among African-Americans and a modest one among whites; the two dominant racial groupings in the county. Using this data, I decided to do a little experiment.
First, I look the 2008 vote for the same-sex marriage ban and reversed it. If a precinct voted 60% in favor of the ban, I estimated that 40% favored same-sex marriage. Now this is an imperfect process because some people voting to reject the ban may have done so on purely legal grounds (a desire for it to not be in the constitution), however, this is the base of support I must use. Then, I took the racial makeup (as a percent) of each precinct and factored in the growth in same-sex marriage support for the respective groups. For example, a precinct 50% African-American would see the 50% multiplied by 16% (the growth in support) — leading to 8%. This is an additional 8% support that exists for same-sex marriage in that precinct. By doing this for all the racial groups, I then add the total new percentage supports to the estimated 2008 support for same-sex marriage. With that I got a whole new set of figures.
The map below shows what the estimated support for same-sex marriage looks like today in Broward County. With the polling taken in 2012 it can be reasonably assumed that figures from a vote that might happen today would be even higher, but for this we will assume the polling has not moved much.
The results are staggering, 57.5% in support of marriage equality; up 10 points from 2008. These figures are just the result of an experiment, but I believe it gives a reasonable look at how a modern vote would take place based on how views on the issue have evolved. The precincts that are still estimated to reject marriage equality are those that heavily supported the ban back in 2008. These precincts largely non-white, and fall under the areas that have heavy percentages of naturalize citizens. However, many other minority precincts will have moved onto the pro marriage equality side by this point.
The map below shows the percent that each precinct would be estimated to grow in support for marriage equality based on the math I laid out before.
The precincts with the largest growth are also the ones that had the largest minority populations. This conforms with the polling shifts being strongest with minority voters. Most precincts would on average gain 7.5% to 10% of the vote in favor of marriage equality.
It cannot be determined for sure how a new vote on same-sex marriage would play out in Broward without polling. It should be universally believed that the county would have rejected the ban in 2012 considering the huge shifts in support among minority voters. The racial minorities and conservative whites that fueled the ban’s support in 2008 are quickly shifting their views as the issue has continued to gain support as time goes on. A vote in Broward could yield similar numbers to my statistical experiment or not. However, for same-sex marriage to clear 60% in Florida it WILL need at least 60% support in counties like Broward. Broward is one county where support for same-sex marriage is likely well above 50% at this point. However, as for clearing 60% statewide, more work will need to be done.