assembling accurate futures crowdsourcing precise contingencies modeling critical forecasts forecasting definitive insights mapping intelligent contingencies mapping the future calculating probable futures predicting calibrated predictions exploring definitive insights formulating contingent futures aggregating predictive forecasts crowdsourcing precise wisdom crowdsourcing calibrated forecasts aggregating predictive wisdom

News

Results of 2020 U.S. Election Risks Survey

By Juan on Oct. 20, 2020, 11 p.m. GMT

Metaculus and Rachel Aicher, Adjunct Lecturer in Political Science at Hunter College, City University of New York, are excited to announce the results of a forecasting session on risks to the 2020 U.S. election! We solicited probabilistic predictions about risks to the U.S. general election on 3 November 2020, including on issues relating to effective electoral administration, potential contestation of results, likelihood of civil unrest, and smoothness of any potential transition. We are building on the important work of others, including the Transition Integrity Project and Edward Foley, who have also made the case that a landslide or otherwise decisive victory for either candidate would minimize the probabilities of such risks. Our aim is for relevant election security decision makers and the general public to be able to make informed choices on the basis of these probability estimates and, if needed, to make necessary preparations.

Below we have included a summary of our results, our methodology, and our analysis. For more, please see our full report.

Summary of Results

Voter intimidation and challenges: Experts’ median prediction for the number of voter intimidation/challenge calls that will be made through the Election Protection nonpartisan voter protection national hotline is 8.4k (80% CI: 1.5k - 18.3k). The number of such calls made during the 2016 election was approximately 4.7k. Experts assigned a 5% probability to over 20k calls being placed, which would be more than four times the number of calls made in 2016.

Presidential election result known: Experts predicted a median of 8 November 2020 (80% CI: 3 November 2020 - 13 January 2021) for when the presidential election result will be called by the Associated Press.

Concession by losing candidate: Experts assigned a median of 40% to the probability of President Trump conceding if the election is called by the Associated Press in favor of Democratic nominee Biden. They assigned a median 95% chance to Democratic nominee Biden conceding if the election is called by the Associated Press in favor of President Trump.

Domestic election interference: Experts’ median prediction for President Trump ordering an agency or personnel under his authority to directly interfere in the vote counting process and for these orders to be successfully carried out is 10%.

Resolution by state legislature or Supreme Court: Experts assigned a median of 7% to the probability of at least one state legislature appointing its own slate of electors instead of assigning electors in strict accordance with ballots cast by their state’s voters. Experts’ median prediction for the Supreme Court issuing a decision that determines the result of the election is 19%.

Election-related violence and unrest: Experts predicted a median of 60 deaths (80% CI: 0 - 912) occurring due to election-related violence between 20 October 2020 and 20 January 2021. Experts assigned an 8% probability to over 1,000 deaths. Experts’ median prediction for President Trump invoking the Insurrection Act between 3 November 2020 and 20 January 2021 is 10%.

Confusion as to who will be the next president on Inauguration Day: Experts assigned a 2% probability to it being unclear who is exercising the powers of the presidency after 12PM EST on 20 January 2021.

Methodology and Logistics

From 8 October to 15 October 2020, predictions were made for 10 questions related to risks to the 2020 U.S. election.

Two groups of experts were asked to participate: (i) subject matter experts (SMEs) and (ii) trained forecasters (TFs). SMEs were defined as political scientists with expertise in American government and electoral systems. They have several years of experience in political science research and have been closely following developments in the lead up to the U.S. general election on 3 November 2020. TFs were defined as the top 1% out of a total pool of approximately 13,000 forecasters here on Metaculus!

A total of 17 experts (8 subject matter experts and 9 trained forecasters) participated and submitted 220 predictions for aggregation into a consensus distribution. During the entire forecasting session, experts could submit multiple predictions for the same question and collaborate via a comment section underneath each question. Experts made 31 comments, sharing analyses and insights with one another across all questions.

Analysis and Implications

A key challenge for policy makers and risk analysts is to identify not only which scenarios are possible, but also which are probable and consequential enough to merit significant attention and resources for planning and mitigation. To this end, this forecasting study aims to map out the landscape of different risks related to the upcoming U.S. presidential election. This research brings together multiple experts’ perspectives to offer timely analysis and insight into election processes, transition issues, and public safety risks highlighted by prior studies and scenario planning. Experts were nearly certain that it will be clear who is president by Inauguration Day (only a median 2% chance otherwise), but were less confident in predicting exactly when the results will be called and noted significant probabilities of vote count interference, violence, and civil unrest. We take a closer look at some of these scenarios below.

Results indicate that experts anticipate higher levels of voter intimidation and challenge complaints than in the previous general election. The median prediction for the number of voter intimidation or challenge calls that will be made through the National Election protection hotline was 8,400 — roughly double the calls made during the 2016 election. It is likely the voting process will be more fraught than usual.

Although the median prediction for when the presidential election result will be called was 8 November 2020, the wide 80% confidence interval ranging from 3 November 2020 to 13 January 2021 reflects a high degree of uncertainty, with some participants commenting that they anticipate delays due to mail-in ballots and vote counting issues.

Experts had divergent expectations of the two candidates’ likelihood of giving a concession speech if the election is called for their opponent: 95% for Biden, yet only 40% for Trump. This suggests a smoother transition if the election is called for Trump, as Biden is highly likely to concede, but a more contested and acrimonious transition if the election is called for Biden, since Trump will probably not offer a traditional and timely concession.

Results indicate it is unlikely authorities under the President’s remit will interfere in the vote counting process, although the experts’ median prediction of 10% likelihood for this scenario still gives cause for concern.

There has been much public conjecture as to how delayed ballot counting and contested results might be ultimately resolved, and results of this expert survey suggest a nearly 1 in 5 chance the Supreme Court will play a decisive role in the election outcome. Participants found it less likely (median 7% probability) that state legislatures will depart from established processes for assigning electors, however.

While large-scale violence and military intervention to quell civil unrest seem unlikely, experts still judged these possibilities to be far from remote. Experts predicted a median of 60 deaths occurring due to election-related violence, with an 80% confidence interval of 0 to 912 fatalities that reflects a high degree of uncertainty. Still, the real possibility of violence is a notable departure from the peaceful transitions that have been the hallmark of past U.S. elections. Results indicate an 8% probability of over 1,000 election-related deaths — suggesting that while widespread sustained clashes are unlikely, this possibility warrants real concern. Experts assigned a 10% median prediction that President Trump will invoke the Insurrection Act to mobilize troops during the transition period.

Overall, expert predictions indicate a non-trivial risk of vote count interference, contested results, and election-related violence — potentially departing from the established norms of concession and the peaceful transition of executive power that have characterized past U.S. elections. With expert predictions indicating the incumbent will likely refuse to concede if the election is not called in his favor and a nearly one-in-five probability the Supreme Court will play a decisive role in determining the election outcome, policy makers would do well to take these issues seriously.