Introduction
Welcome to the
/politics polling discussion thread for the general election. As the election nears, polling of both the national presidential popular vote and important swing states is ramping up, and with both parties effectively deciding on nominees, pollsters can get in the field to start assessing the state of the presidential race. Please use this thread to discuss polling and the general state of the presidential or congressional election. Below, you'll find some of the most recent polls, but this is by no means exhaustive, as well as some links to prognosticators sharing election models.
As always though, polls don't vote, people do. Regardless of whether your candidate is doing well or poorly, democracy only works when people vote, and there are always at least a couple polling misses every cycle, some of which are pretty high profile. If you haven't yet done so, please take some time to
register to vote or check your registration status.
Polls
Below is a collection of recent polling of the US Presidential election. This is likely incomplete and also omits the generic congressional ballot as well as Senate/House/Gubernatorial numbers that may accompany these polls. Please use the discussion space below to discuss any additional polls not covered. Additionally, not all polls are created equal. If this is your first time looking at polls, the
FiveThirtyEight pollster ratings page is a helpful tool to assess historic partisan lean in certain pollsters, as well as their past performance.
With the conclusion of both major parties’ nominating conventions, pollsters scrambled into the field to conduct polls of swing states and the national race. The result has been a slew of high quality pollsters releasing their numbers on Wednesday as well as today, which paint a picture of the electorate right after the candidates are expected to have received a temporary
convention bounce.
Election Predictions
Prognosticators
Prognosticators are folks who make projected electoral maps, often on the strength of educated guesses as well as inside information in some cases from campaigns sharing internals with the teams involved. Below are a few of these prognosticators and their assessment of the state of the race:
- Cook Political Reports - Charlie Cook's race ratings are well regarded in the political field, and he's been in the business for a while. Cook is known to incorporate both public and nonpublic (ie. internal polling) information in his projections. Also covers (and is perhaps better known for) Senate and House races.
- Larry Sabato's Crystal Ball - Larry Sabato has also been in the political prediction business for a while now, and the team at the UVA Center for Politics has been fairly successful in past cycles. Towards the end of the election cycle, Sabato has a policy of making a call in each state, refusing to leave any race in the tossup category. Also covers Senate and House races.
- Inside Elections - Inside Elections, run by Nathan Gonzales, is the successor to Stu Rothenberg's Rothenberg Political Reports, which used to be a part of Roll Call. This prognosticator did fairly well in 2018, though Rothenberg himself boldly (along with some other pundits) predicted in 2016 that Donald Trump's path to the presidency was nonexistent. Also covers Senate, House and Gubernatorial races.
- Daily Kos Elections - Daily Kos Elections is the direct successor to the Swing State Project website, which merged with Daily Kos in 2010. Despite the liberal lean of the site as a whole, the Daily Kos Elections predictions tend to be fairly even-handed, if not even slightly bearish for Democrats. Presidential numbers aren't up yet but they have Senate, House and Gubernatorial races.
- RRH Elections - Red Racing Horses Elections is a site founded by former conservative-leaning members of the Swing State Project community. Despite the conservative nature of the commentary, like DKE, the race predictions tend to be fairly neutral, if slightly bearish for Republicans. Like the Crystal Ball, RRH will call all races before the election so that none are left in the tossup category. The presidential rankings have not been published but they do cover Senate, House, Gubernatorial and Row Officer (statewide elected officials, such as state Attorney General, Secretary of State etc.) races.
- Media predictions - CNN | U.S. News | NPR | POLITICO
Polling Models
Polling models are similar to prognosticators (and often the model authors will act like pundits as well), but tend to be about making "educated guesses" on the state of the election. Generally, the models are structured to take in data such as polls and electoral fundamentals, and make a guess based on research on prior elections as to the state of the race in each state. Below are a few of the more prominent models that are online or expected to be online soon:
- FiveThirtyEight - 538's runs 40,000 simulations of the election and uses that data to come up with a projected winner out of 100 sampled outcomes. It also breaks down the projection to show potential electoral college results, popular vote share, tipping point state likelihood, and it can even be broken down into state-level analytics.
- Princeton Election Consortium - this is the model run by Dr. Sam Wang, a neuroscience professor at Princeton University. This model has run in the past two cycles as well, though Sam Wang famously said he'd eat a bug if Donald Trump won the election because his model predicted no path to victory for the eventual winner of the 2016 election. Also includes projections for Senate and House.
- JHK Forecasts - the earliest model on the scene this cycle. Jack Kersting's model is one of the newer ones this year and also includes projections for Senate and House.
- The Economist - this is the model run by G. Elliott Morris, who previously had a midterm election model under The Crosstab.
- Niskanen Center - Rachel Bitecofer's projection, which only seems to update a couple times a cycle. Part of this has to do with Bitecofer's central argument that there are generally no swing voters, and electoral fundamentals drive the outcome of the election. This was put to the test in the 2018 midterms, where Bitecofer very early on predicted a Democratic pickup of 40-45 seats in the House, which fell about where the election ended up.
- Lean Tossup - a foreign model from Canada. This model did relatively well in the 2019 Canadian election, but this appears to be the first time they've tried forecasting the US Presidential election. Also includes Senate and House projections.
- Primary Model - Helmut Norpoth's model does not actually change with inputs anymore. The outcome it predicts, stated as a 91% chance for President Trump to win reelection, is premised entirely on the primary vote totals in the New Hampshire and South Carolina primaries from earlier this year. Despite being a bit of an outlier against other models, this one was among the only to forecast that then-candidate Donald Trump was a favorite to win election in 2016, and has been applied to correctly predict the winner of 5 of the past 6 presidential elections.
Prediction Markets
Prediction markets are betting markets where people put money on the line to estimate the likelihood of one party winning a seat or state. Most of these markets will also tend to move depending on polling and other socioeconomic factors in the same way that prognosticators and models will work.
Predictit and
Election Betting Odds are prominent in this space, although
RealClearPolitics has an aggregate of other betting sites as well.