“Who says life is fair?” asks one of my favorite actors in one of my favorite movies. And indeed, life — specifically America’s electoral system — isn’t fair for Democrats. Since the end of the Cold War, we’ve won the popular vote in seven of eight presidential elections, yet lost the presidency three times, appointing only three of the current nine justices on the Supreme Court.
The playing field is just as bad in Congress. Though the Senate majority is up for grabs, it shouldn’t be: Democrats consistently win massively more Senate votes than Republicans, but the Senate skews toward geographic representation instead of popular representation, creating a structural barrier for Democrats. Gerrymandering in the House, somewhat alleviated by recent state-by-state court decisions, will again be extremely tilted against Democrats in 2022. Further, the reason Republicans have gerrymandering power in states that Biden won is that those state legislative seats are themselves gerrymandered.
None of this is likely to change anytime soon. We are winning the most votes, yet we are not winning enough elections. So what, given the biased electoral system, should Democrats do to change that?
Here’s the executive summary:
Run on popular ideas. We already do this, and it’s a big reason we consistently win more votes! Our issues — like social security, increased wages, and most healthcare expansion — are very popular. But we need more than just a bare majority to govern effectively.
Keep innovating. In the past twenty years, we have greatly improved our ability to mobilize our voters to cast ballots. This is because we have learned from data and experience, and now have a better understanding of the best ways to get our voters to our polls (eg, plan-making, social pressure) and the best ways to use our volunteers’ valuable time (eg, talking to their friends).
Relational persuasion. Just as we’ve learned to leverage existing relationships between friends and family for turnout, so must we for persuasion. There’s a deep history of relational organizing through the Democratic and progressive infrastructure, but we’ve lost that thread in the Facebook era. Now, Republicans have out-organized us — especially in pockets of Latino communities — and we need to catch up.
Run on Popular Ideas
Analysts like David Shor and a variety of elected Democratic politicians have correctly pointed out something both obvious and occasionally forgotten: Democrats should run on popular ideas.
But I would note that we already do this. Take the part of the electoral process that the Biden campaign had the most control over: his advertisements. The Biden campaign’s most-run ad* promised affordable healthcare. Their most-run attack ad didn’t highlight Trump’s character flaws, but rather his threats to Social Security and Medicare. The campaign translated their policy plans into concrete benefits for Americans struggling in this pandemic-depressed economy.
While the right-wing media takes pleasure in distorting issue positions, Biden’s platform worked. His popular stances, combined with Trump’s horrific handling of COVID (not to mention his character flaws), led to Biden unseating an incumbent by winning more than 80 million votes. The problem isn't that the message didn't appeal to the majority of voters; the problem is that a majority isn't enough.
Keep Innovating: How We Learned To Mobilize
The Democratic/progressive coalition has shown an impressive willingness to develop and adjust political tactics in response to new evidence. These changes may seem incremental. But willingness to adapt over time means that we are constantly innovating and garnering votes on the margin that could make or break an election. Year after year, we learn from experience, pilot creative approaches, and often validate what works best with gold-standard randomized control trials.
The first modern-era randomized controlled trial in electoral politics was an academic study by Gerber and Green in 2000. Their results largely confirmed standard operating procedure: sending a bunch of volunteers to knock on doors of infrequently-voting Democrats is a smart way to boost base turnout.
As practitioners of politics, this study was useful, but we could do better. The Analyst Institute (AI), founded in 2008 to promote shared learning in the progressive community, immediately got to work on empowering volunteers in a way that would help us win elections. AI established best practices and learned how to hone messages delivered at the doors†. Turns out a conversation consisting of “Are you planning on voting?” / “Yep.” / “Okay, thanks, bye” was easily improved upon.
Leveling up the quality of conversation wasn’t enough; we could do better. The most effective mobilization conversations occur near election day, which means volunteers, who are busy and working for free, need something productive to do before October 15th. Up stepped Vote Forward, who developed an efficient way for volunteers to hand-write and hand-address letters,and then proved that this technique generates an excellent number of new votes per hour.
Expanding volunteer opportunities wasn’t enough; we could do better. Recruiting volunteers via paid field organizers working out of rented offices can be costly, and campaigns need to ensure that their field campaigns minimize cost per net vote. Here Mobilize America was a huge help, lowering the cost of volunteer recruitment and shift signups, so that field could compete with TV ads on a cost-efficiency basis.
Still, optimizing the efficiency of list-based voter contact wasn’t enough; we could do better. All of the above tactical innovations increased the effectiveness of strangers (volunteers) talking to strangers (voters). Perhaps not shockingly, we learned that friends talking to friends or family members talking to each other — “relational voter contact” — is much more effective on a per-conversation basis. The progressive community rapidly ingested the evidence and turned 2018 into the “Year of Relational”; we moved fast enough that we were able to act before Republicans could take advantage. The academics have only recently caught up, publishing results that either party could learn from.
But we can still do better. While relational voter contact is fantastic on a per-conversation basis, it can be difficult to scale up the number of conversations with meaningful impact. Enter polling place tripling. The folks at Vote Tripling developed a tactic where campaign volunteers stand outside an early vote site or Election Day polling location—as they normally would to hand out sample ballots—and intercept voters as they leave. Voters are asked to text or call at least three friends or family members, encouraging those friends to follow the voter’s lead by casting a ballot that day.
None of these approaches, by themselves, changes the course of the election. But, cumulatively, they helped Joe Biden garner the largest number of votes a presidential candidate has ever received, by far. Now, we need to take what we’ve learned from relational get-out-the-vote contact and apply it to persuasion.
What’s the matter with Starr County? The need for persuasion
A record number of votes for Joe Biden isn’t good enough. We must keep innovating, or we will fail to keep pace with both cultural changes and Republican tactics, and the other side will win votes we can’t afford to lose. We saw this happen in the Rio Grande Valley in Texas, where the Hispanic vote decisively swung against us. Clinton won Starr County in Texas, which is 96 percent Hispanic, by 60 points in 2016. Biden won it by just 5.
The Wall Street Journal conducted a deep dive into the change of heart those Texans on the border experienced over the past four years. I broke down their explanations on this twitter thread (useful for those without a WSJ subscription), but one common theme is that activists convinced their friends/family.
After Mr Trump was elected, Mr Saenz started Facebook pages spreading pro-Trump messaging to his friends. He helped organize a Rio Grande Valley “Trump Train” of vehicles, and papered a downtown boulevard in Trump signs. He recruited family members. His elderly parents said they were persuaded to vote for Mr. Trump when Mr. Saenz showed them videos of social justice protests in northern cities turning violent.”
This hits at the core of relational persuasion. The basic theory of change is that a member of a less politically-engaged‡ community gets excited about a candidate. They start a social media group—using whatever app their network already uses, usually Facebook or WhatsApp—and start spreading the gospel to their friends and family. This drip, drip, drip exposure to (in this case) pro-Trump propaganda, plus the occasional one-on-one hard sell, can lead to many converted votes.
The Clinton-Trump voters in the WSJ story gave many reasons they voted for Trump this year: his views on the oil industry, abortion, and guns. But Democratic presidential nominees have held liberal positions on those issues going back decades. The reason for the flip this year was that well-connected locals hopped on the Trump bandwagon early and proselytized to their friends and family continuously over the past four years.
Machines and Unions
Democrats used to leverage the small-network approach via two methods: party-run machines and unions. Machines had many unsavory aspects, but they were undoubtedly effective at winning elections. Machines were able to consistently produce so many votes because their local representatives performed very useful services. Here’s a passage about the Chicago machine from the book Don’t Make No Waves, Don’t Back No Losers. The party cared about making peoples’ lives better at the most local level, and those constituents returned the favor with party loyalty.
In addition to machine politics, unions were crucial in leveraging local social networks to help voters understand just how Democratic politicians and their policies would help working families achieve their dreams. Good political science demonstrates how union membership for white workers reduces racial resentment and increases support for redistributive policies. This mechanism meaningfully affects elections. When states pass Right-to-Work laws that diminish unions’ ability to organize, voter turnout drops and so does Democratic vote share.
These effects reverberate today. I analyzed the correlation between union membership (available at the metro area level from Hirsch & MacPherson) and shift in vote share between 2016 and 2020. Even after controlling for education (Biden improved on Clinton’s margin in more educated areas), urbanity (suburbs swung to Biden), and race, union density is still correlated with Biden overperformance. If a metro area had 32% unionization (like Colorado Springs) rather than 2% (like McAllen, TX), Biden’s margin over Trump would have increased by about 2 percentage points.
Indeed, Rep. Ruben Gallego (D-AZ) credits local Arizona unions as being the heart and soul of Latino organizing in that state. These on-the-ground networks give our party a brand that is distinct from thought leaders who are often ahead of the curve (eg, free college, Green New Deal), but can sometimes misstep in their language:
One Modern Day Solution: Facebook and WhatsApp Groups
Unfortunately for the Democratic party, because of Right to Work laws and unfavorable trade deals, union membership has declined precipitously since the 1970s and continues to fall. Unions as a whole, with their ability to provide cross-racial political efficacy, are irreplaceable. But we urgently need to fill one area unions provide—bringing politics into local social networks. The good news is that Republicans in the Rio Grande Valley have shown us one way it can be done; the bad news is that it isn’t easy to replicate.
The first step is to nominate candidates who inspire passion. People on the ground need to be motivated to volunteer their time advocating for our nominees. These candidates need not be ideologues. Trump was no consistent conservative**, and Obama was incredibly inspirational while holding mainline Democratic opinions.
Second, we need to nurture local political networks. We need to direct volunteers to start Facebook and WhatsApp groups†† of their friends — perhaps even leveraging existing groups when appropriate. Many of these volunteer leaders will need both technical and emotional support (as some of their friends may not appreciate the intrusion of politics into their feeds). While campaigns will find it easier to organize in highly-engaged, highly-educated communities, we need to focus our attention on the less-engaged voters who feel abandoned by the system. In some cases, this will involve getting younger voters to talk to their aunts, uncles, and grandparents. In other cases, we need to find well-connected individuals who are the social glue of their local communities.
Finally, campaigns need to produce content that feeds and sustains these groups. Seasoned operatives realized the necessity of recording all of their opponents’ public appearances in the hopes of catching a Macaca or Chickens for Checkups moment. But campaigns also need to record all of their own candidates’ interactions with the public and have good video editors standing by. You never know when you’ll capture an inspiring moment that goes viral. Simply cutting the best part of a speech and passing it on to local groups can help win votes. A side benefit of this approach is that the campaign will quickly learn what content attracts attention and what is too banal to generate engagement.
Additional content can include recommendations for “how to talk with your family about [insert misinformation].” Whether it’s misinformation around socialism, the cost of food, or the pandemic, we need to fight (or even “prebunk”) organically-spread rumors with organically-spread facts. Campaigns should keep in mind that local influencers might be trusted messengers who can be useful when combating misinformation campaigns.
Even with this formula, many newly-formed political groups will fail. The ones that succeed, however, have the potential to snowball into large networks that convert many voters. Campaigns that take the approach have to accept the loss of some control, tolerating risk for reward.
In essence, we need to take what we’ve learned about the power of relational get-out-the-vote contacts, and apply it to the realm of persuasion. None of this will be easy. There is no magic spell that erases the conservative ideology of voters in key geographies. But if we keep learning, innovating, refining, and reaching for the next opportunity, we can win races that the map says we shouldn’t. The last four years have shown how incredibly important this work is; let’s get better at winning votes so there’s no need to rely on a miracle.
After this post was published, it was pointed out that I neglected to mention groups that already engage in relational persuasion. My tweet thread on some of these groups is here.
The author thanks (in alpha order) Christina Coloroso, Jesse Ferguson, Jackie Gran, Hillary Hampton, Matt Lackey, Sean McElwee, Ishanee Parikh, Mike Podhorzer, Jonathan Robinson, Melissa Ryan, Daniel Schlozman, David Shor, Charlotte Swasey, David Winkler, Matt Yglesias, and Avi Zenilman for their assistance in drafting this note.
*Counting one airing of a 60-second ad as twice that of one 30-second-ad airing since the former is on screen for twice as long. The healthcare spot was run over 20,000 times with $11 million behind it.
†Disclaimer: I’m a former Executive Director of the Analyst Institute but these messaging improvements largely occurred prior to my tenure.
‡ Persuasion, especially persuasion at the presidential level, is more likely to occur among low-information voters. The theoretical framework for why is best laid out in John Zaller’s canonical political science book.
** In fact, voters rated Trump closer to their ideology than Clinton.
†† Twitter, being the home to the politically engaged, is intentionally absent from this list.