Evan's Election Night Viewing Survival Guide
NOTE - Updated 10/29/20 5p Eastern, will be updated several times prior to Election Day
Quick Links: AK AZ FL IA GA KY NC NE NV NH ME MI MN OH PA SC TX VA WI
All states are covered below, but above are links to states I'm more closely watching...
Background
A number of factors will make rapidly determining the winner of the presidential election complex. Traditional exit polling will likely be inaccurate / not used (note 1). Additionally, given the vast number of mailed-in ballots, there could be some delays in how quickly results are available in some areas (note 2)
For those who want to know ahead of time what to look for, below is my chronological guide as to what we may know and when we may know it regarding the presidential race and the senate majority. While I’m an unabashedly pro-Biden, pro-Democratic Senate progressive, I’ve tried to make the analysis as non-partisan as possible. I’ve included the # of Electoral votes and 538 status as of 10/26 after each state [a forecast that required Trump to win FL, NC, AZ, and PA to win even though they all currently Lean Biden).
If this is before 6pm on election day, I recommend you read the notes (which I will move to the end by e-day), and skim the hour by hour “run of show”. I’ve tried to highlight what I think will be knowable / talked about during each hour from 7pm to 1am. Feel free to email or comment with ideas on how to make this better. The companion sheet to this document will include early vote totals, where I can get them, and other useful data. I’ll try to turn it into something that can be printed out for ‘keeping score’ at home. And aim to have it completed by mid-day on 11/3. There will also be a “follow-along at home” live version of the sheet for my predictive models of FL, GA, NC, & VA
The electoral college and US geography makes for an interesting election night. I think it’s likely that by 10:30pm we will see Trump 109, Biden 112, uncalled 223 (this assumes that the networks will take longer to call states because of Exit polling issues). Trump could be as high as 166 by 10:30 (if the networks get aggressive and call OH and TX early). Unless all the polls tremendously underestimate Biden support and turnout, it’s unlikely that Biden will be the winner of more than 200 electoral votes before 11pm (CA closes at 11pm and networks are likely to call its 55 votes for him shortly thereafter. For those who get anxious about a Trump win, I recommend adding 78 to Biden’s number throughout the evening to keep things more even (CA, WA, OR, HI) [and 10 to Trump’s (ND, ID, AK)]. However, I think some things could become clear by 7:30pm, and that’s why I’ve created this hour-by-hour guide.
Below are the 2 questions that I think most commentators will be asking throughout the night, and when / how we might be able to answer them:
What type of win / election is this:
Is this a blue wave election, that is fundamentally changing Washington
If NC, FL or GA are able to be predicted / called before 10p for Biden, this will be likely
They NYTimes just announced they will have their excellent "Prediciton Needles" for these 3 states, and those should help provide a lot of context for the night.
Does Biden eke out a victory, but most other items remain unchanged
This likely means that PA, WI, AZ, FL, NC are not called for Biden before 11pm
Trump claims a come-from-behind victory
Either early results show the polls are wildly wrong, or WI, MI, WI have counts that don’t conclusively finish on Tuesday
Will the Senate flip to Democratic control
I tend to believe that this will follow the patterns of the presidential -- early on we’ll see if GA and NC are competitive and if Gideon is leading in ME. That would then suggest IA, MT, AZ and others are within reach
NOTE - 538 chances that I list are as of Monday 10/26 (I’ll be updating this again prior to 11/3)
Cocktail Pairings (updated 10/30/2020)
When I first sent out my election viewing guide, several people asked what the recommended drink pairings were… I reached out to Washington State Rep. Drew Hansen (the person behind all the great bills Jay Inslee took credit for signing -- Tuition-free college, the first state-level net Neutrality, major gun safety improvements, etc.) who happens to be a mixology expert when he’s not passing revolutionary legislation, being a proud father of tweens, or running D&D games. And, he’s a good friend who has a competitive race for the first time in awhile, so if you like his drink pairings (or, more importantly, his terrific legislative accomplishments) I hope you’ll consider donating on his campaign Web site.
For more details on the analytics behind this blog / who I am and why I’m doing this, see Note 3 at the end.
Cocktail Preparations:
Polls close in the Eastern time zone sections of Indiana and Kentucky -- it’s unlikely that there will be any forecasts at this time. This is my least favorite hour to watch on election night - a lot of people trying to make very limited data (other than early voting totals) fit their predictions. At least half of the ‘what to look for’ predictions they make could be wrong, but it’s hard to tell which half…
Note, I’ll likely start tweeting (I’m @evangrossman) by 7 or 8 if there are any useful nuggets of data.
States with 60 EV close, and we will start getting returns from some areas fairly quickly:
Pairing: (Send Trump Back To) Manhattan - recipe belowGeorgia (16, L-Trump) (two senate races, and if Trump (+5 in ‘16 with Johnson getting 3%) is only 1-2 pts ahead of Biden in areas that he won by 5pts, that will suggest the national polls have been correct and it’s unlikely that Trump will win. Calling the two senate races will be a challenge. Ballots are processed when received, so, in theory, the results should not be delayed much by mail in voting.. At bottom of document are data from the three counties(21% of state) that I will primarily be forecasting from...
Ossoff (D)/ Perdue (R)-- last time Perdue ran in the general (2014) it created a runoff with another Republican. This time it’s expected that one candidate will get >50% of the vote.
Warnock (D)/ Loeffler (R)/ Collins (R)-- Special election, question is whether Warnock can get more than 50% of vote in a red state and prevent a run off. Otherwise, he will likely face Kelly Loeffler in a runoff election.
South Carolina (9, L-Trump) Trump won this by 14.2% in ‘16. However, with cities reporting early it may be a little longer before we have a good sense of how things are going. The more interesting race here is the Senate:
Graham (R) / Harrison (D) - Given how solidly R the state is nationally, it will take a lot for Jaime Harrison to win, but he has raised over $60m, and led in a few polls (although 538 gives Graham a 77% chance of winning)
Virginia (13, L-Biden) - Clinton was +5.3 in ‘16 This will be an interesting state to watch given it’s mix of urban and rural, northern and southern voters. Given how much bluer it has been getting each year, A Biden margin below 7% would mean the race will be close nationally. A Biden margin north of 10% would suggest the blue wave might be happening
Vermont (3, L-Biden) - Clinton was +26.4 in ‘16
Indiana (11, L-Trump) - +19 Trump/Pence in ‘16, no senate race, it will be interesting to see where turnout ends up -- 2.7m votes in ‘16, if 2020 is north of 3.5m it is likely coming from urban areas and shows the Dem base is fired up in even a safely red state.
Kentucky (8, L-Trump) - Like SC, the most interesting thing here is the Senate race, and also like SC, if it goes D, it likely won’t be until much later in the night However, this is a state where we will see how much turnout / mail-in voting can change things. KY is a state that doesn’t allow ballots to be processed until election day (see note 2 below), and also allows votes postmarked on 11/3 to be counted for another few days. So if the Senate race is a squeaker, we won’t have final numbers until the weekend, and there could be recounts, court cases, etc.
McConnell (R) / McGrath (D) - 538 gives McConnell a 98% chance of winning (and they only gave Clinton 81% last time).
7:30pm Eastern Time
Another 38 EV come into play (NC, OH, WV) and we may start getting a few bellwether precincts from VA and GA
North Carolina (15, tossup) - Trump won by 3.6% in ‘16 but polls are much stronger for Biden this year than they were for Clinton. Many municipalities have been processing mail-in ballots for weeks (they could start in mid-October), so there should not be delays in getting results (unless there were long lines on e-day or other issues). It will be interesting to see how quickly the networks are willing to call this (see Note 1 on exit polling below). Depending on how quickly some of the more rural areas come in, this may help prognosticators tune their models around early vs. in-person voting, helping guide quicker forecasts for the rest of the evening. At bottom of document are data from the four counties(17% of state) that I will primarily be forecasting from...
Tillis(R) / Cunningham (D) - This is pretty much a must-win for the Dems to take over the Senate and 538 gives Cal Cunnigham a 63% chance.
Ohio (18, L-Trump) - Polling has shown Trump’s base staying with him here more than a lot of other places. Anything short of a 5pt win (he had an 8.1% margin in ‘16) would signal trouble in PA & MI two states that he needs to win (unless he picks up Clinton states in the Southwest). There are a number of still unclear rules around when mail ballots can be processed (it should vary by county, but there could be some statewide changes), so it’s possible that there will be significant delays or that we’ll get results relatively quickly.
West Virginia (5, L-Trump) - Amazing to think that this was one of 10 states Dukakis won in ‘88 that has become a deep red state in the ensuing years (Trump beat Clinton 69-26). I don’t expect that WV will tell us anything about how the national election is turning out. [Also, Dukakis won IA and WI, which both went for Trump in '16]
8pm Eastern Time
By now we should start having some 2016 vs. 2020 comparisons from Georgia, Virginia, and South Carolina that could suggest the contours of the outcome (Blue Tsunami, strong Biden performance, Trump could hold on, something completely unexpected). BUT, it’s unlikely that there will be strong predictions of any Senate upsets or battleground states. Technically, the voting will have ended in states comprising 276 Electoral votes (although in many states, there could still be long lines and voting may continue for a while).
In order of states to watch, here’s what closes at 8p (which means you likely won’t have any good results before 9p). This is when the states with several long-odds (for the Dems) Senate seats will start reporting (TN, MS, AL) as well as ME
Pennsylvania (20, Tossup) - It’s unlikely that we’ll have any significant results out of PA for several hours. Trump won this state by 45k votes in ‘16. PA is one of two close states (WI is the other) that don’t allow processing of ballots before election day (7a in PA, 8a in WI). There is a high likelihood that totals in several large areas won’t be reported until well into Wed. AM. Additionally, because ballots postmarked on 11/3 can still be counted if they arrive over the following 3 days, there will almost definitely be more than 45k votes that have not been counted in the initial vote totals. In a very close election, all eyes will likely turn to PA where a number of likely court battles and other activities could keep the official results from being finalized for weeks. HOWEVER, if some of the red counties get their results in by 9-10p, AND Trump is trailing his 2016 margins, then it’s likely game over.
Maine (4, L-Biden) - This is a complex state for many reasons -- It’s one of two states (Nebraska, closing at 9p is the other) that allocates EVs based on Congressional districts. The statewide and ME-01 are solidly Biden, but Trump may win ME-02; he won by 10pts in ‘16 even though he lost the state by 3 pts, but all indications are the demographics and values have moved on and he may not even win ME-02. Maine historically (especially in the more rural 2nd district) can take a while to get results in, so it’s unlikely we’ll see any calls before 10p. The more interesting race is the Senate:
Collins(R) / Gideon (D) - Susan Collins is the most liberal of the Senate Republicans (kind of like being the tallest building in Kansas), and the voters in Maine seem to have moved on -- 538 is giving challenger Sara Gideon a 62% chance of victory. Collins is likely not gaining more support anywhere, so Gideon needs to be within 5 pts of Biden, especially outside of the cities to win (assuming Biden wins w/ ~55%)
New Hampshire (4, L-Biden) - If Al Gore had gotten 4k more votes in NH, he would not have needed Florida… Since then, the state has become increasingly blue and even though Clinton eked out a .4% Margin in ‘16, polling shows Biden far ahead of Clinton ‘16. Early returns from Manchester and Nashua could indicate if Ds dramatically improved turnout and also percentage wins. If Trump doesn’t win suburban women in NH, he’ll have trouble in other states as well. Meanwhile, on the Senate side,
Shaheen (D) / Messner (R) - Another seat 538 gives a >95% chance of staying with the incumbent
Florida (29, Tossup) - This is a true bellwether state and it’s unlikely that if it’s a very close election we’ll have real results before 9:30 or 10p, despite the ability to start processing ballots 3 weeks before election day. HOWEVER, while counties can not begin reporting results before 8pm, polls close in most of the state at 7pm so results start coming in quickly from some areas. A recent Joe Trippi podcast interviewed Republican strategist Mike Murphy who suggests looking at five counties in FL that typically get results in. I tend to agree with this methodology, and he points out that no Republican has won the Electoral College without Florida in 100 years. The counties to look at (along with links to the county websites for fast election night reporting) are Duval (Jacksonville), Hernando, Sumter, Pinellas (St. Petersburg), and Miami-Dade. If Trump doesn’t exceed his margin over Clinton in these counties, he’s in trouble, or if Biden’s increase in total votes in Miami-Dade exceeds Trump’s increase in total votes in the other counties, Trump is also in trouble. I’ll have these counties summarized in my companion election night sheet.
Here are some numbers for those who want to track along without using the Google sheet that accompanies this…
Alabama (9, Trump) - Trump doesn’t need another 28 point margin to keep the state in the red column, it’s hard to imagine a Biden win. But, in the senate:
Jones (D) / Tuberville (R) - 538 gives Tuberville an 81% chance of beating incumbent Doug Jones (who barely eked out a win in a special election against a pedophile).
Kansas (6, Trump) - Trump had a 20.6pt margin in ‘16.
Marshall (R) / Bollier (D) - 538 gives Barbara Bollier a 25% chance of winning a senate seat in a very red state.
Mississippi (6, Trump) - Hard to imagine a Trump loss here, although the Mike Espy Senate race is worth watching (an Espy win indicates the Dem’s would likely have a 60 vote filibuster-proof majority, assuming Dems win several other very unlikely Senate seats (AK, SC, AL, TX)
Hyde-Smith (R) / Espy (D) - 538 gives Espy a 13% chance, but if the Dem.s can overcome significant voter suppression efforts he has a shot.
Missouri (10, L-Trump) - Trump is up 10pts in the polls and he won the state in ‘16 by 18pts…
Oklahoma (7, Trump) - See Mississippi. OK gave Trump a 36% margin in ‘16. The Senate race appears non-competitive
Inhofe (R) / Broyles (D) - Similar to Markey in MA, 538 gives the incumbent a 99%+ chance of keeping his seat
Tennessee (11, Trump) - See Mississippi. Trump had a 26pt margin in ‘16
Hagerty (R) / Bradshaw (D) - See Oklahoma -- Hagerty favored by 99%
Missouri (10, L-Trump) - Trump is up 10pts in the polls and he won the state in ‘16 by 18pts…
Connecticut (7, L-Biden) - Nothing to see here, no predictive value. A Trump win means that every pollster would have to have been wrong by orders of magnitude
Illinois (20, L-Biden) - See Connecticut.
Delaware (3, L-Biden) - See Connecticut
Maryland (10, L-Biden) - See Connecticut
DC (3, L-Biden) - There’s no scenario outside of outright incredibly incompetent election thievery in which Biden doesn’t win DC (Clinton had a 86pt Margin in ‘16).
Massachusetts (11, L-Biden) - See Connecticut or DC
Rhode Island (4, L-Biden) - See Massachusetts.
8:30pm Eastern Time
Arkansas(6, Trump) - Shortly after the Decision desks call all the 8pm states, they will follow with Arkansas. Which Clinton lost by 27pts in ‘16. In the Senate,
Cotton (R) / Harrington (L) - There are two ways, to paraphrase Edwin Edwards, that Tom Cotton could lose the election to Libertarian Ricky Dale Harrington -- if over the next 6 days he’s caught in bed with a live boy or a dead girl. Other than that, it’s likely the Rs keep the seat.
9:00pm Eastern Time
By now the less cautious networks should have called at least 78 votes for Biden (CT, DE, DC, IL, ME, ME1, MD, MA, NJ, RI, VT) and 82 for Trump (AL, AR, IN, KS, KY, MS, MO, OK, SC, TN, WV) And, Despite 282 votes closed by 8:30, it’s unlikely that 82 will have been called (FL, GA, ME2, NC, NH, OH, PA, VA). Returns from GA, VA, and NC should hopefully be much clearer, and if any Trump16 state gets called for Biden by now (NC could be the first) then we’re looking at a Blue wave. Either way, the GA Senate race will be showing how accurate polling and turnout modeling has been, and give the decision desks (and the rest of us) a better sense of how to weight early returns in other states. Another 12 states (147 votes) closes at 9pm (although, again, there could be long lines in places like NY, MI, WI, TX, etc.) and networks may call some obvious states quickly if their modeling in the 7pm states proved correct. -- LA, NE, NE1, NE3, WY, & SD, ) are likely to be called for Trump before 9:30 in all but a catastrophic loss (either a loss for Trump or a loss of confidence in predictive models by the networks). And, NY will go to Biden (I also think NM will go to Biden early, but there are many scenarios that could make the networks more cautious on calling it given they won’t have other SW states closed before then) leaving several new states undecided - (AZ, CO, MI, MN, NE2, TX, WI).
Turning on your TV set at 9:30 (if you can wait that long) should show Biden 112, Trump 100, and 14 states and 2 CDs still uncalled (211 votes). Thus, if you’re a Biden fan, you could get nervous, but that would be premature (unless MI, PA, WI, VA or another Clinton state goes to Trump, and that’s when you start Googling “how to apply for Canadian passport”)...
Looking at early numbers from Arizona, Colorado and Texas could indicate big wins for the Dems, or show that Trump has come from behind. I think between 9:30 and 10p is when those with access to enough precinct level data can start making accurate predictions on the Senate and Electoral College. We likely won’t hear those predictions until 10:30-11p unless the early returns are overwhelmingly strong in one direction.
Louisiana (8, Trump) - T won by 20pts in ‘16. The senate race could be interesting
Cassidy (R) / several dems - if Cassidy doesn’t get more than 50%, then there will be a runoff.
Wyoming (3, Trump) - Given his 46pt margin in ‘16, anything but a 25pt win shows tremendous softness
Lummis (R) / Ben-David (D) - 538 gives the D’s <1% chance of winning
S. Dakota (3, Trump) - Another likely victory (Trump had a 30pt margin in ‘16)
New York (29, Biden) - In 2016 both Clinton and Trump called NY home and Clinton won by 22.5% (something about those who know him best…) Current polling suggests a similar win by Biden.
Michigan (16, Tossup) - A state that Trump won by .2% in 2016. It’s unclear how quickly returns will come in -- cities can start processing mail-in ballots on Monday 11/2, but they may not all be loaded by the time polls close. This state (as well as PA & WI) should be very good for comparing 2016 to 2020 results (both margin and absolute votes). If Trump loses any of these 3, then he has to win a state that he lost in ‘16. I doubt we’ll have a clear indication of PA before midnight (although we may be able to compare a few cities to see if it’s a Biden rout) but WI and Michigan we may be able to see by 10p,
Wisconsin (10, Tossup) - See Michigan, above. Trump won by .7% out of 3million votes cast in 2016.
Minnesota (10, L-Biden) - Trump only lost by 1.5% in ‘16 so is hoping for a possible pick-up. They typically count ballots quickly and are able to start processing mail-in ballots 2 weeks before election day. The Tina Smith’s Dem. senate seat is unlikely to turn over
Smith (D) / Lewis (R) -
Arizona (11, Tossup) - One of several states the Dems have heavily invested in turning blue. Trump won by 3.6% in ‘16 and he has been behind in almost every poll. The state has extensive experience with mail-in voting, ballots can be processed starting two weeks before eDay, and results will come in fast.
McSally (R) / Kelly (D) - This is a must win for the Dems to flip the senate and current polling shows it within reach (78% likely according to 538)
Colorado (9, L-Biden) - Trump only lost by 5% in ‘16 but polling shows that he is even further behind now. All eyes will be on another “must win” race for the Dems to flip the Senate:
Gardner (R) / Hickenlooper (D) - 538 calculates an over 80% chance that the Dems will pick up this seat.
Nebraska (5, L-Trump) - Actually, NE is solidly Trump (25 pt margin in ‘16) BUT the NE2 CD, around Lincoln and Omaha was won by Obama in 2008, and Clinton only lost by 2pts. Like Maine, NE apportions its EVs by CD (one to the winner of each CD, two to the statewide winner). Electoral college modelers love coming up with scenarios in which NE1 gives a candidate a 270 to 268 win (they love coming up with 269-269 scenarios even more, but I try to eschew doom-modeling). In the Senate another safe R seat is unlikely to changeover:
Sasse (R) / Janicek (D) - Another seat that 538 says is 99% likely to stay with the incumbent
New Mexico (5, L-Biden) - Clinton won by 8.3pts, and Biden is ahead by double digits in most polls. If polling proves accurate in earlier swing states (NC, VA, parts of GA) the networks may decide to call this early.
Luan (D) / Ronchetti (R) - 538 says Dems have only a 6% chance of losing, but if they do it’s hard to see how they flip the Senate
Texas (38, L-Trump) - All eyes will be on Texas long past their 9pm eastern poll close. Long lines in Dallas and Houston, long delays in vote counting in many counties, and other complexities are likely to make the outcome of all races there uncertain until Wed. at the earliest. While they can begin processing ballots when received, getting the results fully tabulated will take awhile. While the Senate seat will only flip in a large wave election, several state legislative races are toss-ups and it’s possible that Rs will lose control of the state House for the first time in awhile (Ds need to flip 9 seats to win).
Cornyn (R) / Hegar (D) - 538 only gives MJ Hegar a 12% chance, but the early voting #s in TX are extremely large and traditional polling could be off.
10:00pm Eastern Time
Another two solidly Trump states (MT & UT) close as does one lean Biden (IA) and a tossup (NV). By now we should have some good 16v20 comparisons for areas from some key 8pm tossups -- FL, NH, ME2. If there is enough data to call Florida for Biden, it’s game over for Trump. If Florida is solid Trump (unlikely this early) then it’s likely we see other must-wins for Biden slipping. Two key senate seats that the Dems are hoping to flip (IA & MT) close now, but it’s unlikely either of those gets called for a few more hours. By 10:30, assuming no new tossups are called for Trump or Biden I expect the EC to be Trump 178, Biden 121, uncalled 145 [Assuming there are no major surprises, folks who are comfortable trusting polls want to I noted earlier, those who are super-nervous about Biden losing should recommend add 78 at this point (CA, WA, OR, WA, HI) and give 10 more to Trump (ND, ID, AK). That would put it at Trump 188, Biden 199.
Nevada (6, Tossup) - Clinton won by 2.4% in ‘16. Modeling based on NM and & AZ may enable the networks to call this before 11.
Iowa (6, L-Trump) - Despite winning by 9.5%, polls show Trump is in trouble and the Senate seat is a chance for a D pick-up
Ernst (R) / Greenfield (D) - 538 gives Teresa Greenfield better than even odds of winning
Montana (3, Trump) - Trump won by 20.4pts in ‘16
Daines (R) / Bullock (D) - Former Gov. Bullock has a 32% chance of flipping the seat according to 538
Utah (6, Trump) - Trump won by 18pts in ‘16
After 11:00pm Eastern Time
A lot could still be unknown by most sane people’s bedtime (on the East coast at least)... Polls in Alaska won’t close until 1am Eastern, and if a Blue wave is sweeping other states, it’s possible that Sen. Dan Sullivan could be in trouble (but we won’t likely know that until 3 or 4am given how long some parts of Alaska take to report in)
When 11 rolls around, several predictable calls will be made as ND, ID, CA, OR, and WA close -- by 11:30, the networks (unless it’s a smash win for Biden) will have Trump 188, Biden 199. After that, it’s a question of how quickly FL, NC, PA, WI, MI, and other states get called. If they’re already called by 11:30 then it’s likely the outcome of the election is clear. It’s also possible Trump will get close to 230 by 11p if OH, TX and GA get called earlier, leaving FL, NC and PA as determinants.
California (55, Biden) - Trump lost by over 4million votes (20pct margin) in ‘16. Expect Biden/Harris to have 5million+ popular vote advantage in CA this time.
Oregon (7, Biden) - Clinton won with a 11pt margin
Merkley (D) / Perkins (R) - 538 gives Merkley >99% chance of retaining his seat
Washington (12, Biden) - Another coastal state that went solidly Clinton (15.7pts)l
Idaho (4, Trump) - Trump won by 31.8%, and is expected to win big again
Risch (R) / Jordan (D) - There have been very few polls, but 538 and others believe Risch is unlikely to lose.
North Dakota (3, Trump) - Only OK, WY and WV gave Trump greater margins than his 35.8pt 2016 margin in one of two states with more time zones than congresspeople
Midnight
Hawaii (4, Biden) - Clinton won by 32.2pt Margin
1am
Alaska (3, Trump) - Trump won by 14.7pt margin, Senate race is likely R, but could go either way
Sullivan (R) / Gross (D) - Current 538 forecast gives native Alaskan Al Gross a 21% chance of beating native Ohioan Dan Sullivan
Bellwether County data:
Rom% = Mitt Romney ‘12 votes/ total votes
Vts = total votes cast
%ofGA = vts/ total statewide votes
Tr% = Trump ‘16 votes / total votes
EV = Early Vote ballots received (as of the updated date below)
EV/16to = EV divided by total ballots cast in 2016 -- ie, Cherokee has already received 85.2% of its total 2016 vote total 5 days prior to the election
EV/state = EV/total EV received in state
Twitter on eNight: [click any link to follow on twitter]
I'm @evangrosmsan Here are some other folks I follow for real-time insights:
@redistrict - Dave Wasserman editor / data wonk at Cook Political
@NateSilver538 - Official feed for 538’s editor in chief
@Nate_cohn - NY TImes Upshot reporter
@joetrippi - Dem campaign consultant
@Davidaxelrod - Former Obama campaign manager, good commentary
@JacobRubashkin - Inside Elections analyst
@maristpoll - Very well respected polling org.
End notes:
(1) Exit polling - In the past, when 70%+ of votes were cast in person, and demographics of those voting in person was not significantly different from absentee ballots, interviewing someone outside a polling place and asking how they voted tended to be very accurate. All current indications show that <50% of ballots will be cast on election day, and that there are some interesting partisan biases toward in-person voting (Trump supporters don’t trust vote by mail as much and polling has shown are more likely to vote in person). Thus most “decision desks” (the network’s statisticians that decide when to “call” a state for a given candidate) will rely on a combination of exit polls (with a lot of complex weighting assumptions) and “bellweather” precincts. For example, in a state like North Carolina which Clinton lost to Trump by 3.6 percentage points, if Biden is up by more than 1-2 points in precincts where Trump won by more than 4 points (a 5-6 point turnaround) that would indicate a likelihood of a Biden win statewide. The decision desks have two conflicting pressures -- A) the need to be right and not make a mistake (super-embarrassing, doesn’t happen often) and B) the need to tell the public what is happening...
(2) Rules around processing / counting of mail ballots vary by state and in some cases by county or municipality. The National Conference of State Legislatures has what appears to be a fairly up-to-date list of rules by state. Here are basic issues -- in many locations, ballots that are received prior to election day can be ‘processed’ in advance. The level of processing varies, in some cases the ballots are opened, unfolded, and fed into the appropriate tabulating machines. In other cases the ballots are sorted and then delivered to appropriate precincts on election day. I do not believe ballots are ever “counted” (in a way that the counter could know the results) prior to election day. In some states, such as Pennsylvania, which do not allow the processing of ballots prior to 7am on election day, this is likely to cause delays. Let’s take Philadelphia and Pittsburgh -- at 7am on Nov. 3, clerks will start processing hundreds of thousands of ballots (here’s a good LA Times article on the process) Additionally, because of a recent PA court ruling, any ballot postmarked on Nov. 3 that arrives within three days (Nov 6) can still be counted. Thus there are a number of scenarios where it’s possible the outcome of the PA election will not be known until the weekend (and in that case there will undoubtedly be further court cases to count more / fewer ballots). Here are the states that don’t allow processing of ballots until election day (and several which allow it ahead of time may still limit it to one or days before). I’ve ranked the list from most likely to be a toss-up to least likely…
Pennsylvania (20)
Wisconsin (10)
Kentucky (McConnell / McGrath Senate race)
Alabama (likely Trump, but the Jones / Tuberville race will be interesting)
West Virginia
Indiana
Mississippi
Wyoming
(3) Who wrote this? Hi I’m Evan Grossman, and have volunteered and worked on elections from an early age (I did lit drops as an 8-year old for McGovern in ‘72). My first job out of college was Director of Analytics for the Dukakis campaign (we came in 2nd). I’ve run election day boiler rooms for multiple elections and have been the person who advises a candidate on when to concede or if a victory is likely. I got it wrong in 2000 (thought Gore would win FL and could win NH) and have been wrong many other times. Usually, if I have access to good data from poll watchers, I can call an election in advance of TV. I doubt I’ll have the time to build the necessary predictive models to do that this time, but will likely focus on modeling the pre 8pm states of GA, VA, & NC to see if they can help predict some national trends. I’ve been volunteering for the Biden campaign in Maine, and plan to spend election night in front of my computer live tweeting interesting statistics from my @evangrossman twitter account.
Sources used:
Poll closing times by time zone and state
Rules for counting mail ballots by state
Google Sheet with latest polling, closing times, Senate race details and 2016 results
https://www.Fivethirtyeight.com -- all of there election forecastsI reached out to Washington State Rep. Drew Hansen (the person behind all the great bills Jay Inslee took credit for signing -- Tuition-free college, the first state-level net Neutrality, major gun safety improvements, etc.) who happens to be a mixology expert when he’s not passing revolutionary legislation, being a proud father of tweens, or running D&D games. And, he’s a good friend who has a competitive race for the first time in awhile, so if you like his drink pairings (or, more importantly, his terrific legislative accomplishments) I hope you’ll consider donating on his campaign Web site.
Prepare in advance: you can prepare now by placing your your gin in the freezer, the vermouth in the fridge, and chilling a martini glass (or two) with ice.”
5pm: Martini
This is a night when we celebrate America (we hope), and in the words of the cocktail historian William Grimes, America is “the land of the brave and the home of the martini.” So let’s start the night off with one of America’s great contributions to global culture. 1.5oz Plymouth Gin 1.5 tsp Dolin vermouth 3 olives / 1 lemon twist (your choice) Stir with ice until very cold; strain into glass; garnish. Back up to 5PM
Comments
Post a Comment