Six states, six factors that decided the US election
19th January 2021
By Andrew Hillman
These words were tweeted two weeks before the US election by Beto O’Rourke, a former Democratic congressman and candidate for the party’s 2020 presidential nomination.
Like many Democrats, O’Rourke was daring to imagine a blue wave sweeping across the nation and handing Joe Biden a transformational landslide victory. Speaking to The Guardian in September, O’Rourke said:
“If Texas comes in for a Democrat for the first time in almost half a century the shock will be seismic. Trump may or may not accept those results, but the rest of the country absolutely will. It will forever reorder what’s possible.”
Optimism among party members that the Democrats could flip traditionally Republican states like Texas had been common for over a decade, ever since Barrack Obama’s first presidential campaign brought millions of new voters to the polls. The idea – often referred to as “demographics is destiny” – was that as the population grew more ethnically diverse and better educated, the Democratic party’s popularity would naturally increase, especially in the youngest and most diverse states like Texas.
But just hours after polls closed on November 3rd, Democrats were no longer dreaming about winning in Texas. Instead, they were worrying that they could lose the presidency altogether.
Trump outperforms predictions in Florida and Ohio, landmark wins for Democrats in Arizona and Georgia
Florida was the first battleground state to report results. Pollsters predicted a 2.5% victory margin for Joe Biden in Florida, so when initial ballots showed the Republican party with a 3% advantage, it suggested that Donald Trump could get the nation-wide polling error required for re-election.
Ohio and Iowa were the next swing states to be declared. Just like 2016, Trump won both states and outperformed pollsters’ predictions.
Trump had a clear advantage among the first votes to be counted in Georgia, Michigan and Wisconsin. However, as more early and mail ballots were counted, Biden edged ahead in all three states.
Arizona counted ballots received before election day first, giving Biden a strong lead. As on-the-day votes were tallied over the next week, Biden’s advantage narrowed, but he hung on to win by just 11,000 votes.
The final state to report was Pennsylvania, considered the key tipping-point state that was most likely to determine the presidency. Trump held an advantage among the first collection of reported votes, but as vote-by-mail ballots were counted the deficit gradually shrank and was then reversed, with Biden winning the state with a victory margin of 1%.
2020 demonstrates the unpredictability of voter attitudes
While Democrat’s dreams of a landslide victory proved illusory, there were hints of O’Rourke’s electoral revolution in the party’s strongest performances.
In 2008, despite Obama winning nearly 10 million more votes nationally than Republican nominee John McCain, the Democratic party were comfortably defeated in both Arizona and Georgia. But in the past decade the party has capitalised on demographic change with the establishment of formidable voter engagement networks in both states and in 2020, these efforts proved enough for victory, by the slimmest of margins. Each triumph was truly historic: Republicans had won all but one presidential election in Georgia since Jimmy Carter left the White House and a remarkable 16 of the past 17 elections in Arizona.
However, other results demonstrated why notions of demographic destiny oversimplify the complex and unpredictable attitudes of voters. Few predicted that Trump would increase his vote share from 2016 in Florida – considered a reliable bellwether state for voting for the winning candidate at all but one election between 1960 and 2016 – and yet lose the election overall. Even fewer would have believed that his victory in Florida would be largely driven by a surge of support from the state’s Hispanic voters.
Changing demographics – combined with Trump’s control on the Republican party – may have quickened the pace at which the electoral landscape is changing, but there is no single formula for anticipating where changes will occur and what they will look like.
Instead, to understand what decided the election in 2020, and is most likely to decide it again in four years’ time, The Debata explored the decisive factors for the six key battleground states. In the timeline below, these states are positioned in square, pulsing boxes – Click on each box to find out why the vote swung in 2020.
Democrats fail to benefit from growing Puerto Rican population
Winning the Latino vote has become increasingly important in Florida as the share of eligible voters who are Hispanic has doubled since 2000.
In 2012, 63% of Latinos voted Democrat as the incumbent president Barack Obama narrowly won the state. But four years later that figure fell to just 58% with many Hispanic voters deserting the Democratic party.
The party hoped to regain that advantage in November, particularly by boosting turnout among the state’s Puerto Rican population, which has grown as many Puerto Ricans have migrated to Florida since Hurricane Maria devastated the island in 2017.While Puerto Rico does not vote in presidential elections, Puerto Ricans are American citizens and therefore gain the right to vote once they move to any US state.
But in the run up to election day there were concerns from Democrats that the party was once again failing to win over Florida’s Hispanic voters.
Florida was the first battleground state to report detailed results and it immediately became clear that Democratic anxieties were well founded. While turnout did increase in many counties with large Puerto Rican populations – including Osceola County, where one in three residents are of Puerto Rican origin, – the Democrat’s advantage in these areas actually fell as voters shifted from Democrat to Republican.
In 2016, 61% of Osceola County voters favoured Hilary Clinton – meaning the Democratic nominee received 35,000 more votes than Trump in the county. But in November, that advantage fell to just 24,000 votes as the county shifted 11 points towards the Republican party.
Trump gains among Miami’s Hispanic voters
But it was a massive swing in Miami-Dade, where two in every three residents are Hispanic, that explains almost all of Trump’s increased victory margin compared to 2016.
In 2016, Republicans won 330,000 ballots in Miami-Dade (34% of the vote) but in November this figure increased by 200,000 – giving Trump a 46% share of the county’s vote.
Miami is home to a large Cuban-American population – a group that tends to be more conservative than Hispanic voters overall. Following the election, Democratic party strategists argued that Biden did not do enough to engage Miami’s Cuban-American population, giving Trump free rein to tie Biden to socialist and communist governments in Latin America.
But Trump’s foreign policy in Latin America – particularly the imposition of tough sanctions on Cuba and Venezuela – also helped him in Miami. Running unchallenged for the Republican nomination may have benefitted Trump too – in 2016 he had to battle through a scrappy primary, including repeatedly clashing with Floridian Senator Marco Rubio, whose parents immigrated from Cuba to Miami in the 1950s.
Record turnout – in both pro-Trump and pro-Biden areas
Texas has voted Republican in every presidential election since 1980In 1976, when Texas last voted Democrat, the electoral alignment was unrecognisable from today – Louisiana, Mississippi and Alabama also voted Democrat, while California, Washington and Oregon were all Republican. but in recent years, Democrats have become increasingly hopeful that the state is on the cusp of a dramatic shift towards the party.
One reason for their optimism is demographic change: three in ten eligible voters in Texas were Hispanic in 2018, up from 22% in 2000. And the state’s median age has fallen to 35, four years below the national average.
Both Hispanic and younger voters tend to lean liberal, but in previous elections in Texas the Democratic party has struggled to persuade them to vote on election day. As a result, in the 2016 presidential election Texas had one of the lowest turnout rates across the country, with only 51% of eligible voters casting a ballot.
So when record early voting numbers were first reported in Texas, it looked like the turnout surge that Democrats hoped for had finally come. 9.7 million people voted in Texas before polling stations even opened on election day. In total, 11.3 million Texans voted in the presidential race – 26% more than in 2016.
But as results were reported across the state, they revealed that increased turnout had not only benefitted Biden. Many counties with the highest increases in turnout – including Comal County, where 43% more ballots were cast than in 2016 – were firmly pro-Trump areas, mostly on the outskirts of major cities.
While Democrats increased their share of the vote in many of these counties, this improvement was counteracted, and in some case dwarfed, by the increased turnout. In Comal County for example, Biden received 28% of the vote – 5 points up from Clinton’s vote share in 2016 – but higher turnout meant Trump’s advantage in the county rose by 7,000 votes.
Massive Trump gains on the southern border
Across the country, the biggest swings in the vote margin from 2016 came in counties on the Texas-Mexico border.
In Hidalgo county, which borders Mexico, just 28% of voters backed Trump in 2016 – by this November his support had risen to 41%.
The red shift was particular large close to the Rio Grande Valley, where Texas, Mexico and the Gulf of Mexico all meet. Trump’s share of the electorate increased by 16 percentage points in urban Webb County and by 24 points in rural Maverick County.
Like in Florida, this sharp shift was driven by Latino voters: in both Hidalgo County and Webb County, for instance, more than nine in every ten residents are Hispanic.
Third-party voters from 2016 prove decisive in 2020
In 2016, Wisconsin was considered so safely Democrat that Hilary Clinton did not even visit the state during campaigning. But Clinton struggled badly among non-college whites, which make up almost six in every ten voters in Wisconsin, and Trump gained a shock victory.
Many voters who defected from the Democratic party did not support Trump in 2016 either, instead favouring a third-party candidate like Libertarian Gary Johnson. In Wisconsin, a state which was decided by just 23,000 votes, there were 190,000 ballots cast for candidates other than Clinton or Trump in 2016. In contrast, just 40,000 voters supported third parties in 2012.
With such small margins separating the two parties in 2020, some suggested that to win the presidency, Biden simply had to persuade 2016’s third-party voters that the stakes were too high for third-party protest votes this time around.
That theory proved overly simplistic: Democratic party victories in Arizona and Georgia would not have been possible without bringing new voters to the polls and convincing some 2016 Trump supporters to switch sides.
But third-party voters did play an important role. With seven in every ten independent voters saying the outcome of the election mattered more than in previous years, the share of ballots cast for third-party candidates plummeted from 6% in 2016 to 2% in 2020.
And in Wisconsin, where the election was once again decided by the finest of margins, third-party voters made all the difference. According to exit poll data from the Associated Press, 92% of 2016’s third-party voters who also voted in November picked one of the two main parties.
Crucially, these voters split decisively for Biden – 63% voted for the Democratic nominee. That gave Biden an extra 30,000 – 40,000 ballotsFor a precise figure we would also need to know exactly how many of those 190,000 voters from 2016 decided to vote again in 2020. If they all did, Biden would gain an extra 45,000 votes; if two-thirds did, he would gain an additional 30,000 votes. over Trump in a state where Biden’s final victory margin was just 20,000 votes.
Signs of a red shift in west Pennsylvania prove illusory
Once Florida reported results on election night – suggesting a Trump win in the sunshine state and a tight election nationally – politicians and reporters all looked for any indications of the parties’ performances in Pennsylvania.
Pennsylvania was anticipated to be the tipping-point state, meaning that if Trump and Biden both came within touching distance of the 270 electoral college votes needed for the presidency, Pennsylvania would be the state most likely to decide the election.
But it was difficult to deduce which candidate held the upper hand in Pennsylvania from early results. Since mail ballots take longer than in-person ballots to process, many states allowed officials to begin processing them before election day, but Pennsylvania’s Republican state legislatures blocked early processing. Because far more Democrats than Republicans voted by mail, it seemed likely that we would have to wait until most mail ballots had been counted – possibly taking days – before the winner in Pennsylvania became apparent.
Instead attention turned to east Ohio, which borders Pennsylvania. The Ohio counties on the state border all showed a 4-6% shift in favour of Trump from 2016. If the shift carried over the border, Republicans would gain enough votes in west Pennsylvania to make a Trump victory likely in the keystone state.
But the pro-Trump shift did not extend into Pennsylvania: it stopped dead on the border.
In Columbiana County in Ohio and neighbouring Beaver County in Pennsylvania, the vote shares were almost identical in 2012 (Republicans received 55% of the vote in Columbiana, 53% in Beaver).
But in 2016, Columbiana – where eight in every ten adults are white without a college degree – shifted sharply towards Trump, with Republicans receiving 69% of the vote. In contrast, in Beaver county – where seven in ten adults are white without a college degree, Republican support rose more gradually to 58%.
This divergence continued in 2020: while Columbiana shifted four percentage points towards Trump, Democrats improved in Beaver County, reducing the Republicans lead slightly.
It was a similar story all along the border: counties in Ohio continued their shift towards Trump, but in slightly more diverse, slightly more educated neighbouring counties in Pennsylvania, the Democrat’s vote share held steady.
Trump improves in Philadelphia but struggles in suburbs
Before the election, few would have imagined Trump gaining support in Philadelphia and yet losing Pennsylvania overall, another surprise outcome from an unpredictable election.
Trump increased his share of the vote in Philadelphia from 15% in 2016 to 18% in 2020. Because turnout was also relatively low in the state capital (cast ballots increased by only 5%, compared to 21% nationally), Trump actually reduced the Democrats net advantage in Philadelphia by 4,000 votes.
However, the Republican party performed poorly in suburban areas. For instance, in Chester County, which narrowly favoured Mitt Romney over Obama in 2012, the Democrats won 58% of the vote.
In total, Democrats gained an additional 105,000 votes compared to 2016 in Chester, Delaware, Montgomery and Bucks – the counties that surround Philadelphia.
Trump, record turnout and demographic change form perfect storm in Arizona
Between 1948 and 2020, Arizona voted Democrat in a presidential election just once.
Maintaining such a streak in a demographically stable state like Louisiana is one thing, but in Arizona, communities and institutions that supported Trump four years ago are entirely transformed from those that voted for Dwight D. Eisenhower in 1952.
In 1950 less than 400,000 people lived in the Phoenix metropolitan area – today it is home to over five million Americans. Arizona’s dramatic population growth has been driven by migration from other states and from Mexico – less than half of Arizona residents were born within the state and in 2018, almost one-quarter of the state’s eligible voters were Hispanic, up from 15% in 2000.
Arizona has historically had low turnout but in November, a record 65% of eligible voters cast a ballot in the presidential race.
This increase mostly aided Biden: Democrats led by 16 points among voters who did not vote four years ago according to exit poll data from the Associated Press. Within Arizona, grassroot movements which emerged a decade ago in response to state laws permitting racial profiling against Hispanic citizens have been credited with boosting turnout, particularly among young Democratic supporters.
But Trump also deserves blame for the Republican party’s defeat in Arizona.
Many of the state’s indigenous residents were angered by Trump’s approval of the Keystone pipeline in 2017, and by his slow response to the Covid-19 pandemic, which has disproportionately impacted indigenous people. Biden’s share of the vote increased by four percentage points from 2016 in both Apache County and Navajo County – home to many of Arizona’s tribal communities.
Trump also repeatedly feuded with former Arizona senator John McCain – including suggesting that the former naval pilot, who was held as a prisoner of war in North Vietnam for six years – was not a war hero as he preferred “people who weren’t captured.” Following McCain’s death his wife, Cindy McCain, declared that she would vote for Biden, as did former Republican Arizona senator Jeff Flake – signifying a mini-revolt against Trump among Arizona’s most prominent Republican figures.
Biden wins battle for the suburbs
Following Trump’s surprise victory in 2016, political strategists began to debate how the Democrats could win back the white voters without college degrees who had the deserted the party since Obama became president in 2008.
One theory was to fight Trump’s right-wing populism with liberal populism by choosing Bernie Sanders as the party’s nominee. Instead, the Democratic party put forward Biden – a more moderate and less divisive political figure.
There was no clearer example of the benefits of that decision than in Georgia.
Democrats held little support among non-college whites in Georgia even before 2016. In 2012, when Obama received 37% of votes from non-college whites nationally, his share among the demographic in Georgia was just 16%.
That same year, just three in every ten white voters with college degrees in Georgia supported Obama. But Trump’s divisive rhetoric and many of his policies have alienated educated white voters – they were against the construction of a wall on the US-Mexico border (39% supported, 57% opposed), and support both a national mask mandate (59% for, 39% against) and re-joining the Paris Climate Agreement (65% for, 30% against).
Biden’s repeated efforts to frame the election as a referendum on decency and unity, rather than focusing on policy, was designed to convince educated white conservatives to vote Democrat. It worked – exit polls suggest that 38% of Georgia’s white college voters picked Biden in November.
The strategy was particularly effective in the Atlanta suburbs – highly educated, predominately white communities that have traditionally voted Republican. For instance, in 2012 the Republican party won 55% of votes in Gwinnett County, worth a county-wide lead of 30,000 ballots. By November, Democrats held the advantage in Gwinnett County with 58% of the vote (worth a lead of 76,000 ballots).
Urban turnout decisive in razor-thin victory
Turnout soared in the 2020 election, with 27 million more votes cast than four years earlier. But many urban counties with large Black populations were among the areas to see only modest increases in turnout.
In Philadelphia, where 44% of the population is Black, there was just a 5% increase in ballots cast; in Bronx County, New York (also 44% Black) turnout rose by only 7%; and in Baltimore (63% Black) turnout actually fell by 1%.
But in Fulton County and DeKalb County in central Atlanta, the number of cast ballots increased by 20%, in line with the national average. The increased turnout in central Atlanta resulted in an additional 81,000 votes for the Democratic party – far more than the 12,000 votes that ultimately separated the two candidates in Georgia.