Here's just a straight average of national polls in years with incumbent presidents on the ballot. President Trump's numbers are worse than Jimmy Carter's at this point during the 1980 presidential campaign, nearly as bad as Gerald Ford's in 76. Not exactly in great company...
Given Biden is up 8 nationally, the only way you can come up w/ a prediction for the election that doesn’t have him favored ~80%+ of the time is to ignore polls or to analyze them incorrectly. Maybe they’ll be wrong, but they’ve been pretty accurate in the past & are a good bet.
Our forecast is maybe even a bit NICE to Trump right now, given that the model thinks Biden will lose ground b/t now and Nov and constrains the possible outcomes to a reasonable range for the polarized era. If we just use polls, his win % could be 90%+.
Polls in 2016 showed that a modest polling error in Trump’s favor could result in his winning the electoral college. That’s what happened! Now, polls show that you’d need a polling error bigger than those in 80-90% of past elections!
Our presidential election forecast finds that Democrats might have to win the national popular vote by 3 percentage points or more to be favored to win the electoral college: projects.economist.com/us-2020-foreca…
I have been trying to process this week’s unrest through the historical lens of the (GOP-driven) politicization of protests against police brutality & the corresponding political importance of (white) racial rage & some things have become very clear to me on what happens next. 1/
2/ Seems like lots of people are falling into Trump’s familiar trap of making the criminal murder of a black man by the police into a radicalized issue of riots and “law-and-order” that will only end up helping him consolidate votes among racist whites and ethnocentrists.
3/ That’s 1 of the reasons why I think we’ve seen community activists stress that long-term progress will come from nonviolent protests. When ppl start arguing about rioters they agree to Trump’s strategic politicization & surrender the high ground (police shouldn’t kill people).
These results are consistent w/ Biden leading Trump nationally by ~8 points, the worst margin for an incumbent president at this point in the election cycle going back at least to the 1940s. Reconcilable only with Biden's odds being much higher than Clinton's at this pt in 2016.
I want to share a bit of polling data with y'all. Over the weekend, I cooked up an average of the last 2 months of polls in every state. Then I trained a model to project polling averages in states that don't yet have any actual polls. The results are here. Preeeeetty blue...
I also programmed a little toy election simulator based just on these projections. It generates national, regional and state-level polling errors given the empirical distribution of error in polls fielded this far before an election. Pretty blue...
#NEW from me: The Economist is now publishing state-level estimates of 2020 vote intention based on my analysis of our our polling data from YouGov. Let me explain how this model works & what it tell us about what might go down in November. THREAD economist.com/united-states/…
The problem: National polls can mislead us about what might happen in POTUS elections. That's b/c there can be large gaps between how left- or right- leaning competitive states are compared to the country as a whole—& the electoral college elects the POTUS, not the popular vote.
State-level polls are also few and far between these days, and many still haven't fixed the problems that caused SOME of them to misfire in 2016 (no education weights being the biggest issues). We think that a lack of weights for political lean could be a similar issue this year.
THREAD Covid-19 is spreading from America's cities to its exurbs & rural heartland. Which areas are likely to face the worst of it?
We modeled the disease's fatality rate & find the South & Appalachia are particularly vulnerable. Policymakers, take note. economist.com/graphic-detail…
Who we think is vulnerable to covid-19 is based on many factors, including their age, underlying health conditions, access to health care and recent exposure to potentially infected people. We acquired data on these indicators as well as case and death counts for every US county.
We're primarily interested in who is susceptible to dying from the virus, rather than just contracting it. This is called the fatality rate, & although we only have data on detected (vs all real) cases, we think the CFR is a good approximation for actual infection fatality rate.
The American far-left refusing to enter partnerships and engage in strategic bartering with their analogous in the dominant two-party system will doom any chance they have to wield actual influence influence over politics and policymaking at a national scale.
And again, I think it’s a deeply unserious attitude that not participating in the political system at all is better than supporting a deeply flawed but proximate candidate in a two-person race. Why sacrifice all your agency when there is still a preferable outcome on the table?
I will say that Sanders and his closest allies seem to understand these tradeoffs. Of course, Sanders has been fighting this battle for 40 years in Congress, so we should expect him to know better than DSA types how to actually achieve political goals.
So far for 2020, handicappers have been laser-focused on how Obama-Trump voters say they're going to vote in Nov. I think they should be equally, if not more, concerned with how moderate Democrats (IE: Romney-Clinton and voters) feel and how that varies by potential Dem nominee.
Spoiler alert... I have the answer, but the data are embargoed ;)
I hear this argument a lot; "Sanders does better with non-voters!" people say, "so he just needs to get them to vote!"
That's, *REALLY* tough to do, so the relevant group is not non-voters but infrequent voters -- people who dropped off b/t 2012 & 16
Throughout the 2020 primary so far our model has been a bit more "conservative" than others (see: RCP, 538) in calculating each candidate's support. Those smoother trends come from a few sources:
First, by design, our aggregate (henceforth called 'the model', as it's really a model of latent voter opinion) needs multiple polls before it decides to shift estimated support for a candidate. This makes it less sensitive to outliers and avoids reacting to noisy data. Why tho?
@YouGovUS We had YouGov ask over 2,000 Democratic voters to rank all the 2020 candidates over the past month. We used those responses to simulate ranked-choice voting in the primary.
How does RCV work, you ask? Just a reminder....
@YouGovUS 0. Count up all the ballots 1. Did someone win a majority of the vote? If yes, they win. If not.... 2. Anyone who listed the last-place candidate gets their votes redistributed to their next-ranked option. 3. Re-tally the votes and repeat 1-2 until someone gets a majority.
#NEW I wrote about "electability" and the 2020 Democratic primary for the paper this week. We combined political science and polling data and found that a moderate Dem with strength among working-class whites has the best chance against Trump.
The 2020 election is likely to be close. Trump's approval rating & GDP growth suggest that he'll win about 49% of the two-party vote, per our modeling. He's clearly competitive, especially when Democrats need to win the popular vote by about ~2.5pts to win the electoral college.
So the "electability" of the candidates is a particular concern for Democrats.
So what do we know about who is more likely to win general elections?