Bernie Sanders lost the 2020 Democratic Primary, in a much more damaging way than he lost in 2016, and there has been a lot of handwringing as to why this happened. This post will be my sorry excuse for an explanation, and also a look into lessons that can be learned from this. The core of the way in which political campaigns attempt to garner votes is by identifying the demographic groups (sorted on the base of geography, race, income etc etc) that can most effectively and efficiently be targeted to secure an electoral majority (you can also read my previous post, which touched on campaign strategy and Cambridge Analytica, to learn about how Big Data has shaken up this formula in recent elections). This is because one of, if not the, most important factors in determining how an individual votes is their social and group identities (this view is laid out much better than I could hope in Democracy for Realists, by Achen and Bartels). On this view, the most pressing question we must ask of Bernie’s 2020 electoral campaign is about the nature of its hypothesised ‘winning coalition’ – who exactly did Bernie envisage coming to vote for him?
The answer is somewhat unconventional, yet also completely unsurprising to anyone who has ever heard Bernie speak (he’s one of those politicians who only really has one speech, which he has given with minor contextual perturbations pretty much every time he’s faced a crowd since he was first elected to national office in the ‘90s). In a two-party system, you can increase the number of votes you receive by taking voters away from your opponent, or by convincing people who do not usually vote to vote for you. His plan was to appeal primarily to this latter group – ‘We need to appeal to young people and disenfranchised working-class people who are giving up on the political process’ and ‘The swing voters that we’re most concerned with are the non-voters to voters’. This would, of course, require a huge increase in voter turnout – ‘the simple truth is we are going to need to have the largest voter turnout in the history of American politics’. This strategy was referred to by Ryan Grim of The Intercept as building a ‘New Electorate’, but in reality reflects an age-old contention of the socialist-influenced ‘progressive’ left – that if a politician appeals to the material interests of the working-class, they will turn out in droves to elect such a candidate. The assumption underlying this strategy is that non-voters are budding progressives who refuse to vote because no candidate is sufficiently progressive to interest them. A concomitant driver of this electoral contention is the phenomenon that Bernie condemns as ‘corporate influence over the Democratic Party’, referring to the idea that Democrat elected officials are in hock to large moneyed interests within Washington D.C., and that this relationship goes a long way to explaining how Democrat politicians betray the will of their voters and pursue economic and social policy that is more right-wing than the actual preferences of those who elected them. According to this theory, were it not for these corporate interests being used to favour centrist Democrats, the ordinary Democrat voter would choose candidates with policy platforms much more similar to that of a Bernie than those of a Joe Biden or Kysten Sinema, as evidenced by strong polling support for policies like free college tuition and Medicare-for-All. This isn’t necessarily a bad contention, and the most promising signs were in 2016, where Bernie managed to beat Hillary in states such as Nebraska, Oklahoma and Michigan by securing the support of working-class whites. The fact that this is the demographic that, in states like Michigan and Pennsylvania, handed the presidency to Trump was met which much fanfare in the aftermath of the 2016 general election, as many on the side of Bernie contended that these voters were tired of the milquetoast neoliberalism of the Corporate Democrats and had thrown themselves at the feet of a populist promising salvation – and that had Bernie been running, the Democrats would have held on to states like Pennsylvania, where Sanders ‘is popular with traditional, working-class, industrial voters’ according to Jeff Weaver, Sanders’ campaign manager in 2016.
The first sign that this theory of politics was shaky outside of the specific context of the 2016 election was… literally the next national election, the 2018 midterms, where all members of the U.S. House of Representatives and one-third of U.S. Senators faced re-election. The midterms were a coup for the Democrats, where they managed to flip 41 seats in the House and overturn the Republican majority. 21 of these flipped seats were in 2016 swing states. Interestingly, it saw a massive increase in turnout by ‘people we’ve historically modelled as neither Democratic nor Republican’. One thing I will have to explain is the factionalism within the Democratic Party itself – it is a big-tent party, which due to the U.S. two-party electoral system has to encompass basically the entirety of ‘the left’ within U.S. electoral politics, and as such there is huge ideological diversity amongst Democrat elected officials, activists and voters. Some of the more ‘left-wing’, Bernie-aligned subgroupings within the Democratic Party are Our Revolution and the Justice Democrats. When the Democrats hold a primary to determine who gets the Democratic Party nomination for any particular office, be it President of the United States or the Railroad Commissioners for the state of Texas, the candidates in these primaries will seek the endorsement of prominent Democrat politicians, elder statespeople and groupings whose ideological position they best represent. Now, back to 2018. Of the 41 House seats flipped by the Democrats, none of them were flipped by a Sanders-aligned candidate. Of the 21 House seats in swing states that were flipped, 16 of them were flipped by Biden-endorsed candidates, and none of them were flipped by a Sanders-endorsed candidate. Regression analysis by the University of Virginia Center for Politics showed that Bernie’s flagship ‘Medicare-for-All’ policy was a voter-loser in 2018. In the aftermath of the 2016 election, much was made of the counties that had voted for the Obama in 2012 but switched to Trump in 2016 – most of these counties were won back by the Democrats in 2018, but not a single one of them by a Sanders-aligned candidate. In the 2020 endorsement game, Democrat Representatives and candidates in swing seats massively favoured Biden, as polling showed he has the best downballot effect in their districts. Jason Hicks has already written excellently about the results of the 2018 midterms in the 2020 swing states, so I will just quote him:
‘One candidate in a potential swing state (Colorado) that Sanders endorsed won—in the relatively safe, already blue 2nd district. Candidates he endorsed in Florida (also endorsed by Our Revolution (OR) and Justice Democrats (JD)), Iowa (also OR and JD), Penn. (JD and OR) and Wisconsin (one in each state) all lost in the general. In Iowa’s third, the primary candidate he endorsed (along with JD & OR) lost to Cindy Axne, who went on to flip the district and has now endorsed Biden. Cindy got 20,000 more votes than the 2016 candidate. Sanders’ Iowa candidate got a little more than 4,000 votes over the 2016 total. His endorsed candidate in PA got 76 more votes than the Democrat in 2016 (yes, 76). Randy Bryce (endorsed by Sanders, JD—and Biden) in Wisconsin got 30,000 more votes than the 2016 election.
The Justice Democrats endorsed two candidates in Arizona—one went on to win a safe seat, the other lost in the primary to Ann Kirkpatrick, who went on to flip the seat, increasing the votes by 25,000, whereas the Justice Democrat in the safe seat (who was an incumbent) got 32,000 fewer votes. Now, one should note in 2016 the seat was uncontested, but a key premise of the Sanders argument is that his approach can win independents and Republicans and new voters that others cannot. In FL’s 27th, their endorsed candidate lost in the primary to Donna Shalala, who went on to flip the seat with Biden’s endorsement. Their candidate in GA failed to flip the 1st, whereas Democrats where able to flip the 6th—in a district that had been rated “safe” for Republicans in 2016 and had a bigger absolute and relative vote deficit in 2014 than the 1st had.
In Michigan’s 11th, JD’s candidate (also endorsed by Brand New Congress (BNC)) lost to Biden-backed Haley Stevens who went on to flip the seat, increasing turnout by nearly 30,000 votes. They had two MI candidates in the general, and neither won.
In the crucial Nebraska’s 2nd Congressional District, the JD-endorsed candidate saw a decline in 16,000 votes from 2016.
In PA, which was one of the three states to give Trump the Electoral College, Dems flipped four House seats—none were from JD-backed candidates. JD had one candidate in the general—the candidate Sanders backed in the 11th that gained 76 votes from 2016. Their candidate lost in the 7th’s primary to Biden-backed Susan Wild, who flipped the seat.
While Democrats were only able to win one House seat in Texas, it was not one of JD’s 4 candidates (three of whom BNC also endorsed) in the general there that did that.
While the Democrats flipped three seats in Virginia, the two OR backed candidates did not succeed in winning their general elections and neither significantly impacted turnout.’
That is a lot to take in, and paints a very grim picture for Sanders. The reason why candidates who were either endorsed by him or were endorsed by organisations which also endorsed him did so poorly is that these candidates were not attracting votes from those that Bernie said they would – the young and the disenfranchised. The Democrats were winning votes from different demographics entirely, which had overwhelmingly voted Republican in 2016 – suburban, high school-educated white women and voters aged above 50, despite the fact that there was markedly increased turnout from people who hadn’t voted in 2016. Remember what I said earlier about electoral strategy – Democrats didn’t need to appeal to the non-voters that Bernie was trying to court, because they were managing to take votes from Republican candidates. This wasn’t necessarily a killer blow for Bernie’s electoral contention – the fact that ‘centrist’ Democrats were winning over people who had voted Republican in 2016 isn’t mutually exclusive with Bernie and Bernie-aligned candidates driving a surge of non-voters, but the fact is that the former was happening to a much greater extent than the latter.
I’m now going to examine exactly why Bernie was wrong in the electoral contentions that I outlined in the first section. Firstly, Bernie’s assumption that people who don’t usually vote are more inclined to his policy views than are people who do vote. The largest ever study of American non-voters, which examined the people who were eligible to vote in the 2016 election but did not do so, was released in 2020 – A great summation of the findings was written by Yascha Mounk. The most striking headline result is that ‘If they all voted in 2020, non-voters would add an almost equal share of votes to Democratic and Republican candidates’. This higher-level result is matched by more specific polling:
Similar results were found by other studies, with non-voters evenly split between Democrat and Republican policy stances, and with a of majority non-voters opposing five of eight flagship ‘progressive’ policies:
The second of Bernie’s premises which I discussed earlier was the contention that Democrat elected officials adopt policy positions that are more right-wing than those which would be supported by Democrat voters, because these elected officials represent the views of Democrat donors, rather than representing the views of working-and-middle class Americans who form the Democrat voter base. Here, he was half right – Democratic Party elected officials’ policy positions and voting records do massively better reflect the policy preferences of wealthy Democrats and donors to the Democratic Party, at both state and federal levels – but this doesn’t make them more right-wing than the average Democrat voter, it makes them considerably more left-wing. That more wealthy and educated Democrats are more left-wing than working Democrats is already shown in those two previous studies, but the most brilliant research into this was done by Berkeley political scientist David Broockman, who found that ‘Democratic donors are somewhat to the left of D voters on economic issues, but *way* to their left on social issues… these patterns hold among the top 1% of donors & in another dataset from Hill & Huber’ and ‘Many worry Democratic donors pull the party to the right on economics. But this data suggests it’s more likely the typical Democratic donor instead mainly pulls the party *to the left on social issues*, perhaps at its electoral expense.’ Bernie’s contention that the average Democrat is a progressive with highly on-trend views about hot social issues, and that the reason these views aren’t represented by Democrat elected officials is because of massive corporate influence, is wrong. How can one square this with widespread popularity of policy positions such as Medicare-for-All, which most Democrat politicians do not support? Political data analyst David Shor gives a good explanation – ‘Democratic elected officials are to the left of 90 to 95 percent of people… the reason is that while voters may have more left-wing views than Joe Biden on a few issues, they don’t have the same consistency across their views. There are like tons of pro-life people who want higher taxes, etc’.
Let’s get into what happened in the primary – Bernie lost very, very, extremely badly. Not because he lost the vote percentage in a landslide (though he did also lose the vote percentage in a landslide – 26.3% to Biden’s 51.8%), but in the sense that his entire electoral strategy was shown to be mistaken. Firstly, youth turnout decreased compared to 2016, and he took a smaller percentage of the vote amongst those young people who did turn out compared to 2016. The best analysis of Super Tuesday voters I’ve seen was a breakdown of the ~1200 counties that voted, done by NPR. Go look over the data yourself if you so wish, but I’ll give a summary here. Across all counties, the more white, employed, college educated, health-insured, and affluent the average voter in a county was, the more likely that county was to favour Bernie over Biden. The more minority ethnic, unemployed, not college educated, lacking in health insurance and poor the average voter in a county was, the more likely that county was to favour Biden over Bernie. This shouldn’t be surprising given what we talked about above – affluent Democrats have more socially and economically liberal policy preferences than blue-collar Democrats. In building an electoral coalition, Biden’s team were massively more realistic and successful, whereas Bernie’s team targeted young people who didn’t turn out for him in sufficient quantities, and targeted a mass of progressive working-class white Democrats and non-voters who, to put it frankly, don’t exist. In the aftermath of Bernie’s romping victories in the Nevada caucus and New Hampshire primary, the internet was awash with pieces predicting that he would triumph on Super Tuesday. A voice of dissent was, of course, Biden himself, who in all his speeches since his dismal showing in Iowa had spoken mainly of one thing – South Carolina, which he referred to as his ‘firewall’. The South Carolina primary was to be held on February 29th, three days before Super Tuesday, and Biden kept telling anyone who would listen that he would win it by such a margin that his momentum going into Super Tuesday would be insurmountable. Why? Because the Biden campaign wasn’t out chasing an illusory or barely participating demographic – he was building up his lead amongst African-American voters, an absolute core of the Democratic Party electorate. Bernie predicted he would triumph on the back of record turnout amongst traditional non-voters, and Biden reckoned he would win off the back of his strong African-American support. What actually happened? Biden beat Bernie in South Carolina by 30 points, and then went on to obliterate Sanders on Super Tuesday. Obliterate is no understatement – He won states like Maine, Minnesota and Massachusetts, which no pollster coming up to Super Tuesday would have predicted. This was the South Carolina firewall coming into effect, with Biden holding a massive advantage amongst voters that decided who to vote for in the few days before Super Tuesday. Bernie had said that he needed a record level of voter turnout to propel him to victory; there was a record turnout in the Democratic primaries, but it massively favoured Joe Biden. In Bernie’s home state of Vermont, there was an increase in turnout compared to 2016, but Bernie only got 51% of the vote, compared to the 86% he had garnered last time around, with Politico writing this amounting to Biden making ‘a mockery of the senator’s (Bernie’s) main argument for his campaign‘. Bernie failed – because while he dreamed of attracting a wave of blue-collar non-voting white Americans, it just didn’t happen. The Sanders campaign was on life-support after Super Tuesday, and the plug was pulled a week later by the Michigan primary. Bernie’s shock victory here in 2016, by a margin of 1.42%, was the biggest upset of the primaries, and was fuelled by working-class white voters. Trump won Michigan by just 0.23% in the general, again fuelled by working-class white voters, so naturally this was one of the examples used by Bernie supporters to advance the view that Sanders would have beaten Trump in 2016 had he run instead of Hillary. Well, in 2020, Joe Biden won in Michigan. He didn’t just win in Michigan state-wide, he beat Bernie in *every single county* in the state of Michigan. He won by a margin of 16.6%. As the link above outlines, this triumph for Biden was on the back of a nearly 33% increase in turnout compared to the 2016 Michigan primary. The funniest and saddest fact from all this is that Hillary didn’t lose to Bernie in places like Michigan because these people loved Bernie’s left-wing stances – regression analysis shows that people who voted for Bernie in the 2016 primaries but then didn’t vote for Hillary in the general were disproportionately likely to describe themselves as ‘conservative’, and fewer than one-fifth of that group voted for Jill Stein (the incredibly left-wing Green Party candidate). Zack Beauchamp, writing for Vox, perhaps put it best –
‘It’s hard to overstate how central the theory of Sanders’s popularity with middle- and lower-income whites was to his campaign and its outside supporters. They saw his unique touch with his voters as not just a strategy for winning the campaign, but a key reason why socialism as a political project was viable in today’s America.’
This supposed popularity was a gross misunderstanding. Bernie winning in Michigan and other similar states in 2016 was not a sign of how much blue-collar white Americans loved him, it was a sign of how much they hated Hillary. As soon as these same blue-collar white Americans got to choose between Biden and Bernie, they abandoned the latter in droves, and by April 1st Joe Biden had won in nearly 83% of the counties which had voted for Bernie in 2016. Another important fact was that Biden took victory in places like Raleigh and Houston, massively outdoing Bernie in support amongst the suburban white women who fuelled the Democratic Party’s takeover of the House in 2018. Biden had virtually no ground game in Super Tuesday states, and in the month leading up to Super Tuesday the Sanders campaign outspent Biden by more than three-to-one, and yet Biden’s intelligent and demographically-aware campaign was untroubled by these disadvantages.
This reflects what is, in my opinion, Bernie’s fatal flaw – he just doesn’t understand American electoral politics. The most unforgivable examples of this were his comments in praise of Castro’s Cuba and subsequent defence of said comments. Florida is going to be a key swing state in the 2020 general election, and at the time of writing, FiveThirtyEight’s election prediction model calculates that there is a 16.8% chance that Florida’s electoral votes will determine the entire election, second only to Pennsylvania in importance. In the 2016 election, Hillary lost Florida by only 1.2% of the vote. In the 2018 Florida gubernatorial election, Democrat Andrew Gillum lost to Ron DeSantis by 0.4%. Floridians of Cuban descent make up 7% of the population, and if you put Fidel Castro up against gonorrhoea in a poll of Cuban Floridians, gonorrhoea would win. Potentially jeopardising the entirety of the election just to make some ‘well, he made the trains run on time’ point in defence of the Castro regime just about sums up Bernie’s galaxy-brained political nous.
Edit: a few people have brought up the fact that Obama also praised the Cuban regime’s improvement of literacy rates as a ‘gotcha’ response to this. At the risk of sounding unkind, I think that response just betrays shallow thinking. Praising the Castro regime is good politics when you’re already President and are attempting to normalise diplomatic relations with Cuba, as Obama was. Praising the Castro regime is dumb when you’re a Democrat primary candidate who needs to win Florida, was Bernie was.
Perhaps the most succinct explanation of just how poorly Bernie’s campaign understood how to win was its explicitly minoritarian strategy – he ran a purposefully rhetorically and politically divisive campaign (staffed with caustic individuals like David Sirota and Nina Turner) centred around securing 30% of the vote, on the assumption that, facing a fractured field, this would secure him a plurality and hence the nomination. His campaign were absolutely shocked when this strategy didn’t work, because moderate candidates dropped out and the vote consolidated around Joe Biden, as this article in The Daily Kos points out. To best sum up how utterly damning this is, I’ll include a twitter exchange between progressive Bernie supporter Jordan Uhl and Youtuber/living meme Joey Saladino from back in February 2020, when Bernie was leading a highly balkanised field:
When the final results of the primary were in, the tally was hilariously close to what that ‘Sanders vs Moderates’ comparison had shown: The moderates had consolidated around Biden, as Saladino had predicted, with his final vote share being 52%; Bernie had failed to convert any supporters and ended up with 26%. If Joey Saladino, a YouTube ‘prankster’ (yes, I feel like a boomer using the term ‘prankster’) and former pizza chef with a high-school education who drinks his own piss on camera (which Uhl’s last reply references) and walks around Staten Island in a swastika armband, better understands how to win a primary than Bernie Sanders and his entire $210 million campaign, then I cannot imagine a candidate that I would be less likely to support than Bernie.
I could go on and on and on and on, but the underlying point I am making should now be clear: Bernie lost because his electoral strategy was centred around chasing a core of voters who didn’t exist in anywhere near the numbers he thought they did, and no amount of activist enthusiasm, ground game or outspending of rivals could overcome this. Joe Biden won because his electoral strategy centred around solidifying his lead amongst core Democrat voter demographics, and he could do this without spending nearly as much as his rivals and with considerably less campaign infrastructure. Bernie thought he would win the white working-class, because the political ideology he is influenced by has always contended that the support of the working class is conditional only on one’s policies being sufficiently left-wing. This is something we hear time and time again, regardless of how deluded it can be – my favourite example is that of the 1983 UK general election, where a dogmatically socialist Labour Party was crushed by Thatcher’s Tories, and the late Labour MP Gerald Kaufman had referred to Labour’s hard-left manifesto as ‘the longest suicide note in history’. Tony Benn, aristocrat-turned-Marxist and doyen of the Labour left, wrote off the defeat as the result of the Labour Party not being socialist enough to capture the votes of the working class.
You can now clearly see what Bernie’s loss has to do with British politics. Labour lost in 2019 because the Tories were much better at targeting working-class white voters in the ‘Red Wall’, and managed to take 43 of these seats, dealing the Labour Party their worst election defeat since 1931. The Tories managed to do this because they had clearly identified the demographic they had to win over – the ‘Workington Man’. The Labour Party is facing a crisis amongst working class voters – The Tories’ lead amongst C2DE voters was 50% larger than their lead amongst ABC1 voters (for those of you unfamiliar with the NRS social grade model, here’s a primer, but ABC1 can be seen as the broad middle class and C2DE as the working and unemployed classes). No one was surprised by this defeat except Corbyn’s inner circle. A few days before the election, the leak of a phone call between Jon Ashworth and a friend of his revealed him to have said ‘it wouldn’t surprise me – for sake of argument – we held Canterbury because of sort of middle-class, Guardian-reading people, but then the Tories take Bolsover off of Labour’. The point I personally knew beyond all doubt that Labour were doomed came a few hours before the exit poll was announced, when I opined in a groupchat that Bolsover would be flipped and a Labour member friend of mine responded along the lines of ‘No, Skinner’s a Brexiteer and his constituents love him – he’ll be re-elected with an increased majority’ (Ashworth was right and my friend wrong – Labour held the traditionally Tory middle-class haven of Canterbury but lost working-class Bolsover). Of course, the issue that drove this was flip was Brexit, and the big question in British politics right now is whether or not the Tories can hold these new seats come the next election, when Brexit will be much less of an issue. Until a few days ago, I was quietly confident that Keir Starmer’s leadership could be enough to reverse these fortunes, especially in light of Johnson’s bungled handling of the coronavirus pandemic. However, I then saw a couple of graphs, and they terrified me:
In terms of social values, the average Labour voter is much closer to the views of the average Conservative MP rather than the average Labour MP (quite surprising, even to me), and is slightly closer to the social views of the average card-carrying Conservative Party member than members of the Labour Party. Never mind just Labour voters – you can see that the little grey dot representing the social views of the average UK voter falls well to the right of the dot representing the social views of the average Tory MP. Of course, this wholly ignores economic policy – but that doesn’t matter, and to explain why I shall go back to David Shor. Because many voters hold ideologically inconsistent views, issue salience in voting decisions hinges massively on issue prominence – basically, voters will align with Party A on some issues and Party B on other issues, and will choose to vote for the party whose position on the most prominent issue of the election aligns with their own. This sort of phenomenon is hinted at in the tweets by Broockman which I shared above, when he points out that Democrat elected officials may be damaging their electoral prospects by voicing the social policy preferences of donors, which fall far to the left of the social policy positions of Democrat voters. This is also why, despite the fact that Red Wall voters align much more closely with Labour economic policy than Tory economic policy on issues like taxation and welfare spending, the Tories managed to secure their votes; the most prominent issues of the election were Brexit and Jeremy Corbyn’s leadership (party leadership is a prominent issue in every election, because the party leader is aiming to become the country’s head of government and is seen as representative of the wider party), and on those issues they aligned much more closely with the Tories, because they hated Jeremy Corbyn and voted overwhelmingly to leave the EU. If the Tories are clever, they will realise that they can keep their support in the Red Wall in the next election by centring social issues, where their policies fall much closer to the average voter than those of the Labour Party. After the focus of politics has moved on from coronavirus, the Tories will need to continue to talk about Brexit as much as possible through the trade agreement negation process, and then engage in a culture war. They can’t rely on the issue of leadership pushing people to their side at the next election, as they have in the last two elections, because Starmer is 45% up on Ed Miliband and 56% up on Corbyn in terms of favourability after 100 days of leadership. We can see this culture war already taking shape – 70% of Brits disagree with the idea of removing the Churchill statue from Parliament Square, and only 18% agree, so the Tories have made a great big show of protecting it and probably threw a small party when Labour-aligned figures such as Owen Jones argued passionately for its toppling. There will be, I suspect, a lot more of this to come – culture war is now a staple electoral strategy of the international right thanks to the influence of Steve Bannon, who I hold to be one of the greatest political minds of the last decade, regardless of how much I disagree with his ideology. Labour need to make sure that they emphasise those parts of their policy platform which fall most into line with the preferences of the voter demographics they wish to target in the next election. Given the utter stranglehold that the SNP have over the Westminster seats in Scotland, Labour will need to win England if it is to regain power, and so the Red Wall will need to be recaptured, Workington Man by Workington Man.