On any given day, my role is to write. On this particular day, I am supposed to write about what 240 million people decided – whether they stood in line or stuck down an envelope and which box they checked on a piece of paper.
But at this particular moment, I have nothing to write. I am watching people talk loudly on a television screen and reading people type in ALL CAPS on Twitter and I am thinking about the importance of not writing.
Journalists, supposedly chastened by a 2016 Trump win that they did not see coming, were going to be very cautious this time around. No more flailing New York Times needle. No more decimal places on FiveThirtyEight’s homepage – instead, a cartoon fox in glasses explaining to you what voters might do. And yet countless articles were still written over the past few months predicting that a Biden win was more likely than a Trump win, and so a lot of people are shouting online and offline because they have been caught by surprise. Because yet again, somehow, they trusted the polls.
Maybe we will still see a President Biden. I understand you are here looking for factual certainties instead of “maybes” and I can’t give you those right now at a presidential level. Here is what I can say:
1. The polls were probably inaccurate. Again.
When I say “polls” here, I’m not talking about the actual process of voting in an election, I am referring to very expensive surveys that ask about a thousand US adults who they will vote for (although it’s confusing that these both have the same name).
Overall, most polls gave Biden a lead nationally. A significant lead of 5 to 10 percentage points appeared as soon as the surveys started and persisted right up until election day. Pollsters claimed that they had learned the lessons from their 2016 mistakes. In fact, Biden was so far out in front, according to the pollsters’ questionnaires with strangers, that even if they were as wrong in 2020 as they were four years ago, Biden would still probably win comfortably.
It’s possible, however, that they were actually more wrong this time around – either because they found it even harder to track down and speak to 1,000 adults who accurately represented 240 million voters, or because Trump voters were even more reluctant this time to tell a stranger their preferred candidate. Or both.
As of around 11.30am EST on Wednesday, it looks like if Biden does win, it will not be by a large margin. Unfortunately, it is going to take a very long time to count all of the votes. Four key battleground states – Pennsylvania, Nevada, Michigan and Georgia – still have tens of thousands of absentee ballots uncounted. Election workers might need days, not hours, to get to a final tally.
2. The electoral college is still broken.
In US presidential elections, there are two kinds of “wins”: by popular vote and by the electoral college. Only the latter means you become president. The two do not necessarily go hand in hand. In 2000 and 2016, Democratic candidates who won the popular vote still did not go to the White House. It is quite possible that Biden will win the popular vote and lose the electoral college.
Currently, it looks like Biden will win the popular vote, and he still has some hope of winning the electoral college, too. Most of the votes left to be counted are absentee or mail-in ballots. We know that those ballots are more likely to be Democrat than Republican. Early voting has long skewed Democrat, but this time there is also a pandemic that is more likely to make Democrats choose to vote by mail over voting in person, compared to Republicans. (Republicans are more likely to have consumed media that has downplayed or flat-out denied Covid-19.)
Take Pennsylvania, for instance, where, at time of writing, 84% of votes have been counted and Biden is losing by more than 300,000 ballots. There are almost 1.1m ballots left to count and Biden is winning three mail-in ballots for every one that goes to Trump. A little simple arithmetic (winning approximately 75% of 1.1 million votes = 825,000) indicates that Biden might comfortably win, once the votes are fully counted.
3. Some women voted for Biden. Some women did not.
Initially, this article was supposed to analyze the demographics of support for each candidate so let me try to at least throw my editor a bone.
As much as I am thoroughly skeptical of polling, exit polls are slightly more accurate – they interview far more people (this Edison survey spoke to 14,318 adults, whereas most polls speak to around 1,000) and they speak to people who have already voted rather than asking people days, sometimes weeks, ahead of time who they maybe plan to maybe vote for.
The exit polls suggest that whereas men were pretty evenly split (49% Biden, 48% Trump), women as a group tended to favor Biden (57% Biden, 42% Trump). But, in case you haven’t noticed, women are a pretty big old group. Demographics cannot be divided into neat slices of the pie – Americans can be rich and Black and male and living in a suburb and college-educated and each of those factors can influence someone’s vote differently.
Despite the fact that Trump described Mexican immigrants as “bad hombres” and “rapists” during his 2016 campaign, Hispanic voters in Florida still showed up for Trump because many Cubans appreciated his anti-communist stance. So I’m not placing much faith in these numbers either. People do not vote in tidy groups.
So I do not have much to write. Except to resurface the article I wrote on 9 November 2016, and make the same plea: avoid predicting the behavior of 240 million people, avoid reading those predictions, avoid complacency.
Journalists will continue to create charts predicting future presidents as long as readers continue to demand them. I do not know how many times polls have to be wrong or how wrong they have to be for us to finally walk away from the dangerous seduction of predicting political outcomes. Seeing graphics that tell us the future is hypnotic, but it is very important to be awake. Especially now.