This year our Presidential forecast model heavily favours Joe Biden to win the 2020 election, in one of the most high turnout elections in American history. Currently leading the national opinion polls by 9 points, and ahead in states with a total of 351 electoral votes, at this stage Biden losing the election would be a significant surprise.
Trump commands a safe lead in 125 electoral votes worth of states (AK, ID, MT, WY, UT, ND, SD, NE, NE-1, NE-3, KS, OK, MO, AR, LA, MS, AL, SC, TN, KY, IN, WV) as compared to Biden’s 216 safe electors (WA, OR, CA, CO, NM, HI, IL, VA, NY, VT, NH, ME, ME-1, MA, RI, CT, NJ, DE, DC, MD). That means that for Biden to win, he needs just 54 electoral votes from the pool of 197 electors across competitive states.
His easiest set of opportunities are in the Upper Midwest, where Trump is 9-10 points behind in Minnesota, Wisconsin, and Michigan, which together have 36 electoral votes. Winning here would require Biden to win back non college-educated white Obama-Trump voters, which by all accounts he appears to be doing. A recent surge in Covid-19 cases in this region (especially Wisconsin), combined with Trump’s terrible approval ratings on the handling of the pandemic, is likely to shore up Biden’s odds in these states.
Biden’s next easiest opportunity is in Pennsylvania with its 20 electoral votes, where he leads by 6-7 points. Pennsylvania is very likely to be the tipping point state this election, playing an absolutely pivotal role. If Biden wins in Minnesota, Wisconsin, Michigan, and Pennsylvania, he wins the Presidency without having to push any further into Republican territory. Our model gives Biden an 88% chance to win The Keystone State, which is a major contributing factor to his lead in our overall forecast.
Although Biden doesn’t strictly need any more targets, in electoral terms it’s generally wise to have backups. If something goes wrong for Biden in Pennsylvania or the Upper Midwest, he could well be rescued by the Sun Belt States. These are Florida, North Carolina, Arizona, and Nevada, which together have 61 electors and represent a more racially diverse cross-section of the population, requiring Biden to excel with groups other than non college-educated white voters. Courtesy of continued strong support from racial minorities as well as improving numbers with older white voters, Biden leads in each of these states by 4-6 points. These states therefore represent a generous cushion for the Biden campaign, and could plausibly decide the election in his favour were he to underperform with white Midwestern voters, but have a strong performance among minority ethnic groups.
Going one step further, Biden has several plausible long shot targets, namely Georgia, Ohio, Iowa, and even Texas. Together these have 78 electoral votes, with Biden and Trump approximately tied in polls in each state. To some extent, states this red have limited relevance in determining the outcome of the election. However, their high number of electors mean that in the event of Biden victory, they will have a significant impact on the margin of victory in the Electoral College, playing a key role in shaping public perception of the outcome of the election, which could shape both major parties post 2020.
Overall, the current state of the race is bleak for President Trump, however the 8% chance of victory he has in our model is not to be disregarded. It’s important to remember that the President continues to have an Electoral College advantage, such that although national polling may have Biden ahead by 9%, he is only 6-7 points ahead in the most likely tipping point state.
Additionally, it’s worth remembering that polling is not an exact science, and that both random and systematic errors are perfectly common. In 2016, pollsters did not weight polls according to education level, resulting in non college-educated voters being underrepresented in samples which led to a 3 point underestimation of Trump, which was even more extreme in several swing states with large non college-educated white populations. In 2018, pollsters did weight by education and were very accurate indeed, so there’s no particular reason to expect a similar error this year as the problem seems to have been adequately addressed. However, it’s entirely possible that there may be a different systematic polling error, a possibility which our model does account for. If there is such an error, there is no particular reason to believe that it will favour Trump instead of Biden, such that a landslide for Biden is an entirely realistic possibility. Furthermore, even if we did see a 2016 style polling error which underestimated Trump by 3 points nationally and more in several swing states, Biden would still win the election. Biden’s lead is simply much larger than Clinton’s was in 2016, so for Trump to win as the result of a polling error we would have to see a truly historical systematic error across the polling industry.
The other mechanism by which Trump could remain President is via successful large scale voter suppression or vote manipulation attempts. Our model explicitly does not account for these, although it does take into account the normal levels of voter suppression which have been present throughout the nation for a long time. It’s very difficult to gauge what the probability of this happening is, but what you certainly can expect is extensive legal battles, delayed vote counts, and an extremely unusual election night. This year it’s going to be sensible to be patient, and voters will not necessarily know who has won the Presidency by the end of the night. Sticking to reliable, nonpartisan news sources and being aware of the fact that enormous amounts of disinformation are present are going to be critical to surviving the next week with our collective sanity intact.
RedvBlue Politics 