RN
Update with two weeks to go!
Not much has changed from last week, and it was a quiet polling week as the major firms are holding off on key swing states until near the end. You’ll probably see a bunch of polls done after the debate Thursday for the final week and a half of the election season.
Over 35 Million people have already voted. That’s over 25% of people that voted in total in the 2016 election. Texas has had more people vote early than voted for Trump in 2016 - a whopping 52.5% of 2016 totals already. The only thing this really tells us is that enthusiasm is up, and turnout will likely be high. You should not be projecting any results from the party registration D lead in early voting this year since Rs are planning on voting on election day itself.
Changes:
- Iowa moves from tossup to Leans Trump
253 Biden to 126 Trump in states not shown below that I consider “guarantees” for each side. (note: I’m including all of Maine for Biden and all of Nebraska for Trump here, with the simplifying assumption that NE-2/ME-2 split 1-1.)
States to watch and recent polling:
Leans Trump (44 votes) There is a pretty solid 5% buffer between these states and the next closest R states, which makes any D flipping beyond these states quite low probability.
-
Texas (38) I debated putting this one into the tossups category, but held off. Polling is about as good as you can get as a D here, but common sense says Texas will go R. Then again, common sense in 2016 said that MI/WI/PA would go D. Enthusiasm for voting is very high in Texas based on early vote totals so far. If you are hoping for a Trump win, that would make me a little nervous.
Oct 18 - 46% - 47% (Biden +1)
Oct 11 - 49% - 47% (Trump +2)
Oct 8 - 49% - 50% (Biden +1) -
Iowa (6) The below polls aren’t from great polling places. NYT does a poll of Iowa this week that should give us a good feel for how it is actually looking, but I’ll place it back in Leans Trump for now.
Oct 13 - 50% - 44% (Trump +6)
Oct 11 - 48% - 47% (Trump +1)
Leans Biden (66 votes) Trump needs to basically win all of these states, except Nevada, to win the election. Biden has the luxury of leading across multiple states where he really only needs one big one to win.
-
Florida (29) It has been a lull in quality pollsters for Florida lately, as many are holding off until the last week or two to get the big swing states. These leads are smaller than before, but still consistently Biden. 32% of 2016 turnout has already voted.
Oct 16 - 47% - 48% (Biden +1)
Oct 15 - 48% - 48% (Even)
Oct 12 - 47% - 49% (Biden +2)
Oct 12 - 45% - 48% (Biden +3)
Oct 12 - 48% - 51% (Biden +3) -
Pennsylvania (20) Same story on lack of quality polls here in the last week. Though the most recent poll is from Rasmussen and has Biden up 3, which seems like a good sign for him. PA early voting lags behind other states though with only 16% of 2016 turnout having already voted.
Oct 19 - 47% - 50% (Biden +3)
Oct 19 - 45% - 49% (Biden +4)
Oct 15 - 46% - 51% (Biden +5) -
North Carolina (15) This still looks good for Biden. 39% of 2016 turnout has already voted now.
Oct 18 - 47% - 51% (Biden +4)
Oct 17 - 48% - 50% (Biden +2)
Oct 14 - 49% - 49% (Even)
Oct 14 - 46% - 51% (Biden +5)
Oct 13 - 42% - 46% (Biden +4) -
Arizona (11) Another week down, another week of Biden holding his lead in this state. 30% of 2016 turnout has already voted now.
Oct 18 - 42% - 47% (Biden +5)
Oct 16 - 47% - 50% (Biden +3)
Oct 14 - 47% - 50% (Biden +3)
Oct 13 - 45% - 50% (Biden +5) -
Nevada (6) One recent poll done by a Republican firm has Biden up 2 but with a bunch of undecideds. I wouldn’t read much into this one poll and would still consider Nevada leaning to Dems.
Oct 11 - 42% - 44% (Biden +2)
Tossups (34 votes) I’d say both of these are 50/50 states. Trump needs them both.
-
Ohio (18) The most recent poll is from Rasmussen, showing Biden up 1. This is going to be close.
Oct 19 - 47% - 48% (Biden +1)
Oct 12 - 47% - 48% (Biden +1)
Oct 11 - 50% - 47% (Trump +3) -
Georgia (16) Another state looking to come down to a photo finish. The below polls are A+, A-, A, B+ from 538. Quality polling showing a tight race. 41% of 2016 turnout has already voted here.
Oct 19 - 45% - 45% (Even)
Oct 19 - 48% - 47% (Trump +1)
Oct 12 - 46% - 48% (Biden +2)
Oct 12 - 44% - 51% (Biden +7)
Current Prediction
Trump: 188 126 Guaranteed + 44 Leans Trump + 18 Tossups (OH)
Biden: 350 253 Guaranteed + 66 Leans Biden + 16 Tossups (GA)
If Biden undid every executive order Trump did on day one, and the was put in a medical coma for four years, that would be a net win for the US.
2 Likes
One week left update!
Let’s talk 2016 and what’s different.
1: At this point in the race, looking at RCP, Clinton was up 2.2 points nationally and falling hard. Biden is up 7.3 on RCP, 5 points better than Clinton.
2: State-level polling. Biden is 3 points higher in polling in MI and WI, and 1.5 points higher in PA. Add-in that there are way fewer undecideds this time around and that polling now accounts for education-level and we shouldn’t see the swings we saw in 2016 in those states - plus even if we saw the same error level from 2016 Trump would lose as he needs an even larger error than before.
3: Clinton was deeply unpopular. She had negative favorability ratings. Biden does not and represents the “lesser of two evils” choice for the majority of non-Republicans this year. Independents and undecideds largely broke for Trump in 2016. This year by all polling they largely favor Biden.
4. Early voting. Biden has maintained a solid lead throughout and we are up to almost 50% of 2016 turnout already voting, locking in much of his advantage. Election day votes will be heavily pro-Trump already, so the ability for something to take away Biden votes in this last week is limited.
Changes:
- Iowa moves from Leans Trump to tossup Of course Iowa gets a bunch of polls right after I posted my last update. Polls favoring Biden.
- Nevada moves from Leans Biden to Guaranteed This change removes the last state that Clinton won from the battlegrounds, meaning Trump is solely playing defense.
- Removing Trafalgar from polling They released their “data” on one of their polls, which they normally keep hidden. Some major red flags - for example, if I asked you what percent of the black vote Trump would be getting, what would you say? He got 8% in 2016. Maybe he could raise that, to 12% or so and still be considered plausible. Trafalgar estimates he will win 29% of the black vote. They didn’t stop there. 30% of registered D voters apparently support Trump, and Trump is winning with the 18-29 year old age group according to them. The head of Trafalgar, when confronted on this, stated that it was the “50-cent effect”. The ‘data’ was then immediately removed from their website, and I doubt we ever see it again.
259 Biden to 126 Trump in states not shown below that I consider “guarantees” for each side. (note: I’m including all of Maine for Biden and all of Nebraska for Trump here, with the simplifying assumption that NE-2/ME-2 split 1-1.)
States to watch and recent polling:
Leans Trump (38 votes)
-
Texas (38) NYT just released a poll with Trump up 4 in Texas - however that same poll showed that Biden was up 7 based on those who have already voted. Election day voters heavily favor Trump, and Biden isn’t doing great with some of the Hispanic population of Texas. Bloomberg has decided to do an ad blitz in Texas this last week to try and sway late voters to Biden’s side (or just discourage R’s from voting). Still leans Trump, but this state is in play.
Oct 25 - 47% - 43% (Trump +4)
Oct 25 - 48% - 49% (Biden +1)
Oct 20 - 50% - 45% (Trump +5)
Oct 20 - 47% - 48% (Biden +1)
Oct 20 - 45% - 48% (Biden +3)
Leans Biden (75 votes) This is why Trump has such a low chance to win. Pick any one of these states. Any one. And Biden wins.
-
Florida (29) Biden +2 is the story of the pollers recently, but nobody but Rasumussen and Trafalgar has had Trump leading in this state for over a month
Oct 25 - 48% - 50% (Biden +2)
Oct 23 - 48% - 50% (Biden +2)
Oct 22 - 47% - 49% (Biden +2)
Oct 22 - 45% - 48% (Biden +3)
Oct 21 - 50% - 46% (Trump +4) Rasmussen
Oct 20 - 46% - 50% (Biden +4)
Oct 20 - 45% - 52% (Biden +7)
Oct 20 - 47% - 51% (Biden +4) -
Pennsylvania (20) PA is going to look really weird on election night, since they count in-person election day first and then will add mail-in afterwards. Expect Trump to be up big and then the lead to get chipped away through the late night / early morning. This state is close to being “guaranteed”, but as it may decide the election I’ll leave it on this list.
Oct 26 - 45% - 52% (Biden +7)
Oct 26 - 45% - 51% (Biden +6)
Oct 23 - 44% - 51% (Biden +7)
Oct 22 - 46% - 51% (Biden +5)
Oct 20 - 45% - 50% (Biden +5)
Oct 20 - 43% - 53% (Biden +10) -
North Carolina (15) Surprisingly not a ton of polling in NC lately. All non-Rasmussen polls are holding around +3 to +4 for Biden though
Oct 23 - 47% - 51% (Biden +4)
Oct 21 - 49% - 48% (Trump +1) Rasmussen
Oct 20 - 47% - 50% (Biden +3)
Oct 19 - 44% - 48% (Biden +4)
Oct 18 - 47% - 51% (Biden +4) -
Arizona (11) These are all low quality pollsters, and I wouldn’t put much stock in any of them. It seems like the big boys are holding off on this swing state until the last week.
Oct 22 - 46% - 46% (Even)
Oct 21 - 46% - 50% (Biden +4)
Oct 20 - 48% - 47% (Trump +1)
Oct 19 - 46% - 47% (Biden +1)
Oct 19 - 46% - 48% (Biden +2)
Tossups (40 votes) These are all coin-flip states.
-
Ohio (18) No recent polls. Bloomberg’s internal polling showed this as a state that could be won by Biden, so he is throwing $15M at it in ads for the last week. Seems a tossup, leaning more to the R-side right now.
Oct 20 - 48% - 45% (Trump +3)
Oct 20 - 49% - 47% (Trump +2)
Oct 19 - 47% - 48% (Biden +1) -
Georgia (16) Two outliers, one on each side, and a bunch of Even polling. Georgia is maintaining it’s trajectory as the closest state this cycle.
Oct 26 - 46% - 51% (Biden +5)
Oct 23 - 49% - 49% (Even)
Oct 23 - 46% - 46% (Even)
Oct 21 - 49% - 45% (Trump +4)
Oct 20 - 48% - 48% (Even)
Oct 19 - 45% - 45% (Even) -
Iowa (6) We finally got quality polling in Iowa! The +3 and +4 are from A+ rated pollsters, and one of the evens is from an A- rated pollster on 538. Iowa is a true tossup right now.
Oct 21 - 48% - 48% (Even)
Oct 21 - 47% - 47% (Even)
Oct 20 - 43% - 46% (Biden +3)
Oct 19 - 47% - 51% (Biden +4)
Current Prediction
Trump: 182 126 Guaranteed + 38 Leans Trump + 18 Tossups (OH)
Biden: 356 259 Guaranteed + 75 Leans Biden + 22 Tossups (GA, IA)
I’ve been wondering how far off polling will be this year. I assume that the voting distribution is changing somewhat with how much more voting is happening this year. Not sure there are good stats currently about who all of those new voters are.
I have to imagine that likely voter models are a nightmare this year. It’s impossible to predict who will be voting.
For polls, they ask the people how likely they are to vote. The good ones like NYT / Monmouth use probability weights for how likely people are to vote based on their responses, and someone saying they already voted is weighted the same as someone who says they are definitely going to vote (otherwise with the skew to D of early voting, you would end up getting D weighted responses). NYT gives a small weight to those saying they aren’t going to vote, since some actually do (and is the reason behind their larger number of undecideds compared to other polling shops).
Based on people’s responses, they are estimating a very high turnout year - about 150-160 million compared to 135 million in 2016. This high turnout is then reflected in the polls.
Not that the national averages are reflective of what will happen on a state-by-state basis, but you can see the national shifts in voter turnout / who they favor based on national polling from that graphic.
NC just swung red on my map.
Current prediction: 290 D - 248 R
There are people in our neighborhood trying to collect ballots, but they aren’t hitting every house, which is creepy af. I hope no one (regardless of politics) hands over their ballots to them. Since our house has registered Rs, Ds, and independents, I don’t know which of the affiliations made them come to our door.
I am not a fan of ballot harvesting in general. I’m sure it is based on party registration though and in states where it is legal I am guessing both sides are doing something like that.
And for those following along, a promising alternative to PA for Biden is AZ+NE-02. NE-02 looks extremely safe for Biden right now, and while AZ margin isn’t as big as PA, it seems more likely to flip to D at this point- and AZ’s polling errors in the past couple elections have underestimated the D vote, opposite of the PA polling error direction from recent history.
Based on what?
Even acknowledging the outlier status of the WI +17 poll, tons of fantastic polls across multiple states from multiple pollsters for Biden today.
FL +3, AZ +5, AZ +4, AZ +6, MI +7, MI +8, WI +17, WI +5, PA +5, NC +3, GA +4, National +11 YouGov, National +12 CNN.
I still feel pretty good about my 356-182 prediction.
One question I was considering recently as we see record turnout possibilities - does trump get more votes than 4 years ago? I haven’t heard of many people that are voting for him now that didn’t then, but lots that voted for him that regret it.
With the increase in early voting and VBM, what do you think about releasing results as votes are counted? Should early votes be tallied as received, or should counting wait until election day? We are trained to watch results until late into the night for tight races, should that continue with certain states allowing ballot receipt after election day?
Wait until election day to release results, but count early and release as of the closing of polls.
Agree with Bing. Don’t release any counts until polls close, but count and process upon receipt (the Colorado approach)
He has made some inroads with the Latino population (a term I hate because of how diverse that group is) and higher turnout among non college educated white could result in more raw votes for him.