I’m writing a book about polls and democracy
Political polls are not just tools for predicting elections. Used properly, they could empower the people and improve self-government
I’m very excited to announce that I am writing a book with W. W. Norton about the history and future of polls and their indispensable role in a modern democracy.
Here’s a little blurb about it:
Data journalist for the Economist, G Elliott Morris’s NO MARGIN FOR ERROR, a concise examination of political polling — its history, influence, successes, failures, and future — and why it is a crucial tool for a healthy democracy.
The book is due to be published in late 2021. I’m grateful to my agent Lisa Adams at The Garamond Agency and my editor Quynh Do at Norton for believing in this project and teaming up to publish it. I’ll be posting updates here and at my personal website as the writing process continues and as publication (and preorder!) dates approach.
I had the original idea for this book sometime in the summer of 2019. Then, I had just moved to London for a short stint at The Economist’s home base and had a lot of extra time on my hands (summer nights are rather long at that latitude). I spent much of it reading in local bookstores and beer gardens and daydreaming about democracy and the public polling industry.
It occurred to me that the public polling industry has not gotten its just treatment in popular media recently. Although the 20th century brought both the emergence of scientific polling and with it plenty of debate about how they could be used to better democracy, this idea has fallen out of our popular vernacular about political surveys. The conversation about them now is dominated by horse-race coverage and references to forecasting sites such as FiveThirtyEight.
This attention on polls as tools for electoral handicapping has earned them plenty of negative media attention recently. It is commonly understood that polls “missed” the 2016 election. In fact, the New York Times went so far as to claim that data “failed them” in (not) calling the 2016 election for Donald Trump. Polls are similarly under attack in Britain and Australia, where the media have crafted narratives about high-profiles “misses” destroying the industry.
But these narratives about polls are misguided—and maybe even dangerous.
Misguided because the reality is that the 2016 polling data suggested that there was a much higher chance of Donald Trump’s victory than did the shoe-leather reporting by publications such as the New York Times, which was at a loss to explain how it missed the mark so widely. Rather than relying too much on the polls, the media erred by not relying on them enough—and by incorrectly interpreting the certainty of the polling data when it reported on it.
Dangerous because the hue and cry over polls misses the mark. It’s usually not the polls that are wrong but the pundits. Polls themselves are quite accurate, especially by historical standards. People just don’t understand that a polling prediction is like a weather forecast. Told that there’s a 20% chance of rain, no one should be surprised that they’re caught in a sudden downpour.
Instead of calling winners and losers in a binary fashion (proclaiming that Clinton will win and Trump will lose in 2016, for example), polls calculate a range of possibilities for any given percentage (such as saying that Clinton is favored to beat Trump, but she could just as easily win by 5 percentage points as lose by 1).
But more important than ruling out specific electoral scenarios, polls also guide interpretation of public support for government action and policies. Particularly, because majorities are integral to American political thought, legislators are unlikely to act on polls that show diverging opinions—support versus opposition to public health research, for example—within margins of error of each other and are more likely to accept information that reflects real differences among the public.
Condemning polls as worthless is dangerous to this cause. Anyone who believes in democracy must acknowledge that polls play a critical role in the political process. They provide a voice for citizens, serving as a pipeline from the governed to the government and as a bulwark against special interests and despots. Crucially, polls give the people the ability to reign in their leaders between elections. As America faces its worst public health and economic crisis in a century, such a role for polls is on display; the president’s campaign manager, Brad Parscale is using polls to persuade the president against certain actions that could endanger the people (and his re-election prospects).
Polls wield massive influence over key functions of our electoral, judicial and governing systems. They are used in matters as important as determining who gets to run in presidential primaries. In 2016, the Republican National Committee used polls to select candidates for their prime-time debates, elevating the status of those who had large followings and committing to obscurity those that did not. For their 2020 presidential primaries, the DNC followed suit. Importantly, polling thresholds forced several qualified candidates to drop out of both contests. Who knows how different the world would be if Carly Fiorina, the former CEO of Hewlett Packard, or Colorado Senator Michael Bennet had enjoyed actual shots at the presidency in 2016 and 2020?
Public opinion polls have also been instrumental in shaping Supreme Court decisions. Scholars argue that a surge in public support for equality pushed key judges to rule in favor of legalizing same-sex marriage in 2015. Pollsters regularly cite this example as a reason for their industry’s importance beyond the electoral horse-race.
Yet there is no question that the 2016 election revealed that the traditional polling techniques are no longer adequate. The choices pollsters make as they adjust to changes in technology, rising costs, and crises of trust and reliability will determine the very future of the industry—and American democracy. I plan to spend a big chunk of the book discussing what the future of the industry looks like now that the old tools look doomed or impractical.
The problems of the 2016 polling season aren’t new, but pollsters get flack for being on the wrong side of the contests no matter how “close” they are. Their failings were further compounded by a new force in political journalism: election forecasters. When used correctly, prediction models provide context for how likely a candidate is to win an election–given all the polls that have been released during a campaign. But prognosticators can treat polls as being straight on measures of political behavior and that can lead them to overstate a candidate’s lead.
. . .
My argument is that American democracy today relies on polls. Lawmakers look to political polling as the key barometer of public sentiment. When the US House of Representatives was considering impeaching Donald Trump in late 2019, for example, the Democratic Party claimed that the public was behind them, and that persuaded them to take action to sanction the president. Public opinion was a key factor in the fight for same-sex marriage.
Polls thus embolden agents of change. Equally important is their ability to reign in despots. Although Donald Trump believes that a majority of Americans approve of his job as president—they do not, to be sure—his actions are nevertheless constrained by staffers and counsel who seek to avoid upsetting voters more than they already have.
Yet because we distrust polls, especially when they concern the electoral horse-race, they are at risk. Fewer and fewer Americans put their trust in pollsters to accurately reflect the will of the people. Media pollsters have suffered especially hard. Over the past hundred years, accusations of bias, manipulation and misrepresentation against news outlets have left many firms impacted by an angry and rancorous public. Restoring trust in the industry is a key step in preserving polls for the next hundred.
That’s it for now! I’m looking forward to the next 6 months of writing and re-writing (and re-writing and re-writing…) and am excited about sharing the stories that make up the past, present and future of public polling with you all. It’s a long way off, but when pre-orders are available you all will be the first to know.
UPDATE January 12, 2022: The first bound galley copies of Strength In Numbers have been sent off to key contacts. The draft table of contents for Strength is:
Introduction
Chapter 1: Democracy and the Public Will
Chapter 2: Polling Comes of Age
Chapter 3: Machine Politics
Chapter 4: One Bad Apple Doesn’t Spoil the Bushel
Chapter 5: America in Aggregate
Chapter 6: Big Data and Black Boxes
Chapter 7: Taking the Pulse of the Pulse of Democracy
Conclusion