With the Trumperdämmerung upon us, Hillary Clinton is looking to expand the map. Ahead nationally by double digits in some polls, her campaign—and many a political pundit—now believes it can flip a few traditionally red states into her column. On Monday, the campaign announced it would pump $2 million into its Arizona operation and send top surrogates, from Michelle Obama to Bernie Sanders, to the desert. Meanwhile, the YouGov polling model has Clinton pulling into a narrow lead in Georgia, one of only three states it brands a tossup.

It’s easy to see why Democrats think they have a shot in these two states in particular. Arizona and Georgia have long hovered tantalizingly just out of Democrats’ reach. In 2008, Barack Obama received 45.1 percent of the vote in Arizona and 47.0 percent in Georgia, barely missing out on what then would have amounted to a combined 25 electoral votes. That total has since grown to 27 on the heels of transformative population growth, especially among minorities. Since the 2000 Census, Arizona’s Hispanic population has mushroomed by 52.6 percent; in Georgia, it has been the African-American population that has increased by 30.1 percent since 2000.

The demographic shifts have caught the eye of many analysts, some of whom now mention Arizona and Georgia in the same breath as emerging swing states. There’s just one problem: they’re not there yet. White voters still make up a strong majority of the electorate in both states, and it’s currently mathematically impossible for minority voters to hand Democrats a win by themselves.

Take Georgia. The 2014 electorate in the Peach State was 63.5 percent non-Hispanic white, 28.7 percent African-American, and 7.8 percent other voters—yet Democrats Michelle Nunn and Jason Carter both lost their races for Senate and governor, respectively, by eight points. If Clinton performs the same as Nunn did in 2014 exit polls by race, and Donald Trump performs the same as Republican Senator David Perdue, the 2016 electorate would need to be something like 55 percent white, 33 percent African-American, and 12 percent other races in order for Clinton to prevail.

A Winning Electorate for Hillary Clinton in Georgia


Vote Share
Non-Hispanic White (55%)
African-American (33%)
Other* (12%)
Overall (100%)
 
23% 74%
92% 7%
54% 50%
49.5% 49.0%

  Dem Performance, based on 2014 electorate  GOP Performance, based on 2014 electorate

* Calculated indirectly; numbers may not sum to 100%.

Although the Georgia electorate has gotten steadily less white over the past few decades, this would be an eye-popping shift. Georgia’s most diverse electorate ever materialized in 2012, when President Barack Obama also lost the state by eight points despite a racial breakdown of 61.4 percent white, 29.9 percent African-American, and 8.7 percent other. According to voter-registration statistics, the share of white voters has only dropped by about two percentage points since then, while the share of other voters has risen by 2.2 points. (The proportion of African-Americans has slightly decreased, by 0.2 points.) Minority growth alone thus probably will only account for a Clinton improvement of about two points over Nunn or Obama.

Georgia's Likely Electorate in 2016


Vote Share
Non-Hispanic White (59%)
African-American (30%)
Other* (11%)
Overall (100%)
 
23% 74%
92% 7%
54% 50%
47.1% 51.3%

  Dem Performance, based on 2014 electorate  GOP Performance, based on 2014 electorate

* Calculated indirectly; numbers may not sum to 100%.

In Arizona, it’s a similar story. Holding Democratic and Republican performances at the same levels as 2012 presidential exit polls, the electorate would have to be something like 64 percent non-Hispanic white, 26 percent Hispanic, and 10 percent other races in order for Clinton to win.

A Winning Electorate for Hillary Clinton in Arizona


Vote Share
Non-Hispanic White (64%)
Hispanic (26%)
Other* (10%)
Overall (100%)
 
32% 66%
74% 25%
95% 4%
49.2% 49.1%

  Dem Performance, based on 2012 electorate  GOP Performance, based on 2012 electorate

* Calculated indirectly; numbers may not sum to 100%.

At first glance, this seems like a plausible racial composition: the latest Census Bureau estimates put Arizona’s population at 56.9 percent non-Hispanic white, 30.1 percent Hispanic, and 13.0 percent other races. However, Arizona’s citizen voting-age population (CVAP)—in other words, those eligible to vote—is 67.6 percent non-Hispanic white, 20.5 percent Hispanic, and 11.9 percent other races. Furthermore, even eligible minorities are far less likely to be registered to vote and actually turn out than white Arizonans. In 2012, Arizona’s CVAP was 69.1 percent non-Hispanic white, but exit polls pegged the presidential electorate at 74 percent white.

Trends in both CVAP and exit-poll data imply that we should expect Arizona’s electorate to be only a couple points less white and more Hispanic in 2016 than in 2012. A better demographic projection would expect Clinton to improve only slightly upon Obama’s performance in Arizona.

Arizona's Likely Electorate in 2016


Vote Share
Non-Hispanic White (73%)
Hispanic (20%)
Other* (7%)
Overall (100%)
 
32% 66%
74% 25%
95% 4%
44.8% 53.5%

  Dem Performance, based on 2012 electorate  GOP Performance, based on 2012 electorate

* Calculated indirectly; numbers may not sum to 100%.

Of course, the Clinton campaign is not only relying on these states’ diversification; it is counting on boosting Democratic performance in comparison to past years as well, especially with the widely disliked Trump on the other ticket. The problem is that voters in Arizona and Georgia haven’t shown themselves to be very persuadable. No matter the election, no matter how many resources Democrats like Richard Carmona (the party’s U.S. Senate candidate in Arizona in 2012) and Nunn have poured into these states, their white voters have remained intractably Republican, and the Democrats have seen diminishing returns among minority voters. (With non-white support already exceeding three to one, it’s entirely possible that Democrats have maxed out their performance among minorities in these states.)

In Georgia, Democrats appear stuck on 23 percent of the white vote. According to exit polls, John Kerry in 2004, Obama in 2008, Nunn in 2014, and Carter in 2014 all received exactly that amount of white support. African-American support for those Democrats hovered between 88 percent and 92 percent, with the exception of Obama’s astounding 98 percent. In order to win Georgia, then, Clinton must either soften the enmity that white Georgians have harbored against Democrats for over a decade; exceed the African-American performance of the first African-American president; and/or dramatically improve among the mysterious bloc of “other” voters, who are mostly Hispanic, but are also by definition difficult for campaigns to target and turn out. Either way, it will require a feat of persuasion that is virtually without precedent.

Three Winning Scenarios for Clinton in Georgia


Vote Share
Non-Hispanic White (59%)
African-American (30%)
Other* (11%)
Overall (100%)
Scenario 1
27% 70%
92% 7%
54% 46%
49.5% 48.5%

  Dem Performance  GOP Performance

* Calculated indirectly; numbers may not sum to 100%.


Vote Share
Non-Hispanic White (59%)
African-American (30%)
Other* (11%)
Overall (100%)
Scenario 2
23% 74%
99%
54% 46%
49.3% 49.0%

  Dem Performance  GOP Performance

* Calculated indirectly; numbers may not sum to 100%.


Vote Share
Non-Hispanic White (59%)
African-American (30%)
Other* (11%)
Overall (100%)
Scenario 3
23% 74%
92% 7%
72% 28%
49.1% 48.8%

  Dem Performance  GOP Performance

* Calculated indirectly; numbers may not sum to 100%.

Arizona voters have shown more variation in their voting preferences since 2004, when 41 percent of white voters pulled the lever for Kerry and 43 percent of Hispanics did so for George W. Bush. However, the passage in 2010 of Senate Bill 1070, Arizona’s infamous anti-immigrant law that requires police to inquire into the immigration status of those arrested or detained when there is a “reasonable suspicion” they’re undocumented, has starkly polarized the state along racial lines. In the three close elections exit-polled since then, Democratic performance among whites has been between 32 percent and 36 percent, while Democratic performance among Hispanics has been between 71 percent and 74 percent. Only if Clinton can cobble together Democrats’ best showings in those elections—Carmona’s 36 percent of white voters, Obama’s 74 percent of Hispanic voters, and Obama’s extrapolated 95 percent of other voters—will she be able to win Arizona by even one-tenth of a percentage point.

A Winning Scenario for Hillary Clinton in Arizona


Vote Share
Non-Hispanic White (73%)
Hispanic (20%)
Other* (7%)
Overall (100%)
 
36% 58%
74% 25%
95% 4%
47.7% 47.6%

  Dem Performance  GOP Performance

* Calculated indirectly; numbers may not sum to 100%.

It seems clear that an outright majority is out of the question for Clinton in both states. However, there is a scenario in which Clinton’s 45–47 percent in Arizona and 47–49 percent in Georgia is nevertheless sufficient to score a win. Third-party candidates Gary Johnson and Evan McMullin may help Clinton nowhere more than in these two states. In Georgia, Trump could easily bleed enough white voters that he could slip into a danger zone: 68 percent white support, coupled with 5 percent African-American support and 30 percent other support, would drag him down to 45 percent of the total electorate, for instance. And it’s not hard to see plenty of the 6 percent of Arizonans who are Mormon writing in McMullin, depriving Trump of just enough votes to lose him the state. Trump could plunge to 45 percent in the Grand Canyon State with 55 percent of the white vote, 20 percent of the Hispanic vote, and 10 percent of the remaining vote.

But those independent candidacies cut both ways; Clinton is also apt to lose a point or two to Jill Stein, especially in Arizona (she is not on the ballot in Georgia), as well as to Johnson, who may be hurting Clinton among millennials. Indeed, head-to-head polls in both Arizona and Georgia differ little in their winning margins from polls that include third-party candidates. And so we find ourselves back at the two-way exit poll data that indicate Trump has small, but relatively safe, leads.

Ultimately, it is white voters who will determine whether Arizona and Georgia are truly battlegrounds. And if Clinton does win Arizona or Georgia, Republicans can blame the unique third-party dynamics of this election—not the usual narrative of demographic change.