The US presidential election could be closer than the polls suggest (if we can trust them this time)

With less than two months to go until the US presidential election, Democratic candidate Joe Biden leads incumbent Donald Trump in most opinion polls.

But poll-based election forecasts have proven problematic before. The polls were widely maligned after the 2016 election because Trump won the election when most polls said he wouldn’t.

What went wrong with the polls in 2016? And is the polls to be believed this time, or as in 2016, the polls substantially underestimate Trump’s support?

READ MORE:
* 2020 US elections: Bill Clinton, Barack Obama and Uncle Joe Biden, who both survived.
* Why Trump Won’t Win Again in November
* Why would-be Democratic presidential candidate Joe Biden has a lot to smile about
* Only four states will decide if Donald Trump is reelected president of the United States in 2020

Swinging states decide the US presidential election.

The US presidential elections are two-stage races, state by state.

States are assigned delegates roughly proportional to their populations, with 538 delegates in all. The votes of the Americans then decide who wins the delegates in the Electoral College.

In almost every state, the candidate with the highest number of votes takes all the delegates from that state. The candidate who obtains the majority (270 or more) of the Electoral College wins the election.

For the fifth time in US history, the 2016 elections produced a mismatch between the national popular vote and the Electoral College result. Hillary Clinton, the Democratic candidate, won nearly 2.9 million more votes than Trump, but still lost the election.

Trump won states efficiently, by very narrow margins in some cases, converting 46% of all votes cast to 56.5% of the Electoral College. By contrast, Clinton’s huge popular vote recount was concentrated in large states like California and New York.

For this reason, election analysts focus less on national polls and more on “swing states.”

These are states that have oscillated between parties in recent presidential elections (for example, Michigan, Wisconsin, Pennsylvania, Florida, and North Carolina), or that could be about to change (Arizona, Texas, Georgia, and Minnesota).

These states, even just two or three of them, will decide the 2020 elections.

State polls currently point to a Biden win

As part of a large research project at the Center for American Studies, we have collected data from survey averages across all swing states going back 120 days before the election and compared it to the same time periods. in 2016. Our goal was to provide a key benchmark for reading surveys more correctly in 2020.

Charts for all states of oscillation can be found here.

Our research shows that Biden currently has poll leaders in several states that elected Barack Obama in 2008 and / or 2012 and then leaned toward Trump in 2016, including Florida, Michigan, North Carolina, Pennsylvania, and Wisconsin. These five states are worth 90 electoral college votes.

Biden also leads the polls in Arizona, which consistently votes Republican (11 electoral votes).

So if we take the recent polls in these states at face value, Trump would lose 101 of the Electoral College votes he won in 2016 and be completely defeated.

Why were state polls underperforming in 2016?

But state polls were heavily criticized in 2016 for underestimating Trump’s support, as these graphs highlight in our research.

Final 2016 poll averages underestimated Trump’s margin over Clinton by more than five points in several changing states: North Carolina (5.3), Iowa (5.7), Minnesota (5.7), Ohio (6.9) and Wisconsin (7.2). ). This is calculated by taking the difference between the official election result and the average of the polls on the eve of the election.

A review of the 2016 surveys by the American Association for Public Opinion Research examined a number of hypotheses about the bias of the statewide surveys in 2016.

Two predominant factors made the difference:

1. An unusually large number of late decisions strongly favored Trump

The number of “undecided” in 2016 was more than double than in previous elections. Of these, a disproportionate number voted for Trump.

But 2020 polls to date reveal far fewer undecided voters, suggesting that this source of polling error will not be as great in this year’s election.

2. Changes in electoral participation

In 2016, Trump successfully mobilized white voters who are becoming a smaller portion of the American electorate and generally have low voter turnout rates. These were mostly non-urban voters and those with lower levels of education.

This year will likely see high levels of participation from both parties, and potentially an increase in participation not seen in decades, which could further undermine the accuracy of electoral polls.

Reinterpreting the 2020 surveys

The latest state poll averages imply that Biden will easily win the election with an Electoral College victory from 334 to 204.

But if the 2020 polls are as flawed as they were in 2016, then Biden’s current poll results in New Hampshire, North Carolina and Wisconsin are misleading. If Biden loses these three states, the Electoral College result will be 305-233, still a comfortable victory for Biden.

In recent weeks, however, we’ve seen that Biden’s survey results in Pennsylvania and Florida are smaller than the corresponding survey error in those states from 2016.

If Trump wins these two large states (plus New Hampshire, North Carolina and Wisconsin), and the other 2016 results are replicated elsewhere, then he will win the election by a slim margin with 282 Electoral College votes.

Given the statistical range of poll errors observed in 2016, and assuming they are repeated in 2020, current polls imply that Trump has about a one in three chance of winning reelection.

What other factors will come into play?

The Covid-19 pandemic and the controversies surrounding the administration of the elections could further jeopardize the validity of the 2020 polls. Official statistics already show that many voters are trying to make use of voting by mail or in person, early voting.

Access to these alternative forms of voting varies greatly in the United States, so the political consequences are difficult to anticipate.

Trump and his Republican supporters have raised questions about the validity and security of the vote by mail. A recent opinion poll showed that Democrats are far more likely to rely on vote by mail compared to Republicans (72 percent to 22 percent).

Unsurprisingly, Democrats and other groups are filing numerous lawsuits to help ensure that vote-by-mail remains a widely available voting method.

It is highly likely that the courts will be asked to rule on the validity of the results after the election, on the grounds that mail ballots have been incorrectly included or excluded from official counts.

So will it be closer than expected?

For one thing, this year’s elections appear to have historically low levels of undecided voters, a factor that should make polls more accurate. But offsetting this is tremendous uncertainty about the turnout and the votes that will be cast and counted.

All of this suggests that great caution should be exercised when relying on polls to forecast elections. These forecasts are almost certainly too confident.

The other main takeaway: Trump’s chances of reelection are likely higher than the polls we’ve seen to date suggest.

The charts in this article were originally created by Zoe Meers, a former data visualization analyst at the University of Sydney Center for US Studies.

Simon Jackman is the Executive Director of the Center for US Studies at the University of Sydney

This article is republished from The Conversation under a Creative Commons license. Read the original article.