We only have a few facts with which to work if we want to predict the outcome of today's election. I know that I could quadruple my page views by dropping little bits of good data for Trump or bad data for Harris, but the truth is that we don't have enough good data to do more than make educated guesses about the outcome right now. This is not the 1984 election. 

That means that anybody making a solid prediction about the outcome is reading their "gut," the tea leaves, and casting chicken bones to come up with their prediction. 

That's not to say that there aren't data points that matter and that give us solid hints; it's just that without more solid data any prediction is more an assessment of the probabilities based on limited data. Think of these data as helpful but not dispositive. Predicting the outcome of the election is like predicting the winner of the Super Bowl midseason. 

Knowing this, I am glued to Twitter/X just like you probably are, or perhaps you are watching your favorite news channel to get every hint of what is happening. It's human nature. So let's take a quick look at what data makes sense to watch and what doesn't. 

First things first: ignore the exit polls. They are utterly worthless as a predictive tool, and aren't even meant to be used that way. Exit polls are less reliable than public polls, which, as Nate Silver and others have pointed out, are rigged to make the pollster look good, but not right. Pollsters are looking to ensure their rankings in the aftermath are acceptable, not get things right. In a close election, it is impossible to come up with a number that nails the results because the data is nonrandom, and even a perfect random sample still has a margin of error. In a poll with +/- 3, a result of +3 means the same thing as -3. 

Exit polls are worse because they don't even TRY to correct for the nonrandom sample. Democrats are more likely to respond, and the results will show Democrats with a higher vote percentage than the final number. That's reality, and even exit pollsters will say that. If Democrats are 16% more likely to respond to pollsters in pre-election polls, how can a non-corrected exit poll come close to being accurate?

The function of exit polls is to collect data about voter concerns, distribution of voters by demographic, and other data that will be useful well after the election to suss out why a candidate won or lost. They tell you nothing about the outcome. They will be CORRECTED after the election to match the outcome. That is how they work. 

So ignore exit polls. They are not useful data. 

So what is? 

• Early voting numbers tell you something, but given the changes in voting patterns you should be cautious about reading too much into them. With that said, Republicans have reasons to be very happy with the totals, and Democrats are very nervous because their presumed voters have not turned out. 
• Where the candidates went over the past few days. Kamala Harris is obviously worried about the "Blue Wall" because winning them is the ONLY way she can pull this off, barring a miracle. While Trump is expanding the map, Kamala is furiously trying to hold her key states. 
• Reports about turnout in heavily Red or heavily Blue districts. This is anecdata, not real data, but if you aggregate it it isn't nothing. If there are long lines of men in Red areas and only a trickle of voters in Blue areas or the opposite, this is more than a chicken bone if less than solid data. 
• Results from Virginia, North Carolina, and Georgia. Georgia is a likely Trump victory, but margins will tell you something, and if he loses there it will bode ill for his chances. North Carolina is predicted to be close, and if Trump wins with a comfortable margin it bodes ill for Harris. 
• Results in New Hampshire. New Hampshire will almost certainly have no impact in the Electoral College, but it might serve as a canary in the coal mine. If Trump really is in a position to come close or win in New Hampshire that is a solid sign he is well positioned to win. He doesn't need the EVs, but if New Hampshire really is competitive it is hard to see how Harris is doing well in Blue areas. 

The first results will start coming in early, but don't get too depressed or excited until you see a big trend. Your blood pressure won't thank you if you do. 

7:30 PM: NC polls close. They usually also report pretty fast. Some key counties to watch:
- New Hanover: Swingy, trending Dem but Trump did make some gains here in 2020
- Granville: Big swing towards R here in 2020.
- Nash: East of Raleigh, also very swingy

— Alicia Smith (@Alicia_Smith19) November 5, 2024

7 pm: VA, FL, and GA close

GA will be reporting its votes very fast, almost FL-level fast. First readings from a swing state on what may be coming.

VA and FL arent being contested, but FL margin as well as key VA congressional races (VA 02 and 07) could be telling

— Alicia Smith (@Alicia_Smith19) November 5, 2024

(cont) PA will be slower to count due to lack of pre-canvassing,but counties like Allegheny (Pitt)will drop a huge batch right when polls close
Key counties:
- Northampton and Eerie both bellwethers
-Lackawana (Scranton): Harris needs to stay above Clinton number here

— Alicia Smith (@Alicia_Smith19) November 5, 2024

Right now, things look better for Trump than Harris, but with margins so close a minor shift in expected behavior or turnout could shift things in Harris' direction. She doesn't need a huge swing in her direction for the Electoral College to make a major shift in her direction. 

The math doesn’t work for Kamala. She’s down millions of urban and women early votes while rural voters are early voting in greater numbers than in 2020. @CNN @thelauracoates pic.twitter.com/iqCg8xOwAr

— Marc Lotter (@marc_lotter) November 5, 2024

Trump could blow it out--after all, Nate Cohn of The New York Times has said that Trump voters are much less likely to talk to him than Harris voters, so they have to correct for Republican nonresponse in their polls. That "shy Trump" vote may be huge, or the corrections may be good or even overestimate Trump. But the nonresponse bias is real and huge. Ed wrote about this yesterday. 

Four years ago, the polls were thought to underestimate Mr. Trump because of nonresponse bias — in which his supporters were less likely to take surveys than demographically similar Biden supporters.

It’s hard to measure nonresponse bias — after all, we couldn’t reach these demographically similar voters — but one measure I track from time to time is the proportion of Democrats or Republicans who respond to a survey, after considering other factors.

Across these final polls, white Democrats were 16 percent likelier to respond than white Republicans. That’s a larger disparity than our earlier polls this year, and it’s not much better than our final polls in 2020 — even with the pandemic over. It raises the possibility that the polls could underestimate Mr. Trump yet again.

Sixteen points. That is a huge percentage difference in response bias. 

But however well things look for Trump, the ballot counts are what matter. You and I will likely stress out and become elated many times today. 

So sit back, hold tight, and remember that in the end all we have is spotty data, chicken bones, and tea leaves.