Read Post Opinions' live discussion about our election simulator — and 2020 election by the numbers (Jan. 9)

Jan 09, 2020

Read about David Byler's Post Opinions Simulator and the 2020 Democratic primary model that powers it. This powerful simulator gives readers the opportunity to explore what might happen in each primary contest, using a rigorous, empirical model to help them understand what’s possible, what’s unlikely and why.

Read this chat transcript for a deep dive on numbers and the upcoming presidential election, and an exploration of what we can know – and what we can’t.

Read David’s columns here.

Follow David on Twitter here.

Hi! I'm David Byler -- welcome to my first live chat. Today we released the Post Opinions Simulator -- a tool that lets you run your own election scenarios for Iowa (and eventually) beyond! So let's talk about the the horse race, polling, the primary and any other election questions that are on your mind! 

Wasn't there an idea that no matter what happened, George W. Bush was sort of Teflon? He had something like 90% approval ratings in later part of 2001 and it remained relatively high and stable even during bad news cycle after bad news cycle another until his response (or lack of response) to Hurricane Katrina and then it was the 2006 midterms were Republicans lost in places they never expected. Mention it only to point out maybe the idea that Trump's supporter is never, ever going to go down because it hasn't before despite many bad news cycles is a bit premature to say with absolute certainty?

This is a great question -- you're right that George W. Bush's approval rating changed *enormously* over the course of his presidency. You can see it in the Gallup data here. And I think if Trump faced the same sort of crises that Bush faced, his approval might dive as well. 

So far no foreign policy event of the Trump Administration has dominated the news and damaged Trump in the way that Iraq did for Bush. And the economy has generally been good for Trump, while Bush faced a massive recession. I tend to think that if Trump faced a recession or had one bad story dominate headlines (instead of a fast moving train of stories that range from bad to just strange) he'd be in a worse position 

There is an idea the Trump support is low and also rock solid, but it has dipped right? Like during the 30-day government shutdown over refusal to fund his "Border Wall" it went significant down, right?

This is another great point -- I've actually tracked some of the highs and lows in Trump's approval rating. And while his baseline is low, the ups and downs are directly related to events where his competence or policy chops were called into question. The border wall, losing the fight on healthcare -- if Trump does something unpopular on policy, he loses people. If he's just up to his usual shenanigans, his approval rating stays in a pretty narrow band. 

I should point out a lot of White voters aren't supporting his campaign, but it seems like the standard response from that Indiana mayor and his campaign is his next-to-nothing polling among non-White voters is low name I.D., but after months and months of news coverage as well big TV and internet ad buys, can he still low name I.D. to what seems zero growth in support among non-White voters?

There are a couple not-mutually-exclusive reasons why Buttigieg might be having trouble with non-white voters. One is low name I.D. as his campaign often points out. There's something to that -- it's still early for most voters, and he might gain support with non-white voters if he notches some wins in early states.


Another is that he's competing with Joe Biden -- a man who has spent a lifetime in politics and has the halo of the Obama presidency around him. Many non-white voters, especially African-American voters, tend to vote pragmatically. So if Mayor Pete doesn't look like the candidate who can best a) win and b) further their policy goals then he's not going to be their first choice. There are other reasons, but those are some that come off the top of my head 

Seemed like every national outlet called the U.S. Senate election in Arizona rather definitely for Republican nominee Martha Sally and it wasn't until days later that Krysten Sinema was shown to be the winner. Side note is that it will never not be weird to me at least that Martha McSally got appointed to the U.S. Senate immediately after losing although there was a theory that she was very gracious in her defeat and didn't continue to contest the results because she knew she'd get the appointment. We'll eventually find how Arizonans vote in November, but are journalists aware that mail-in ballot means if Arizona is a close state, Trump-Pence will probably look like the winners on election night, but still end up losing Arizona by 1% or 2% so they ought to be really, really hesitant to call it?

I can't speak for other journalists, but I'm aware of this sort of thing ;) 

Things like "Russia, if you're listening, I hope you're able to find the 30,000 emails that are missing"?

This is what makes politics hard to model statistically. 

When we talk about politics, it's not some sort of test tube and petri dish thing where we can use physical laws to predict what will and won't happen. It's a whole mess of people acting collectively, and often unpredictably. There's no way to write a math model that predicted all the twists and turns of Trump and Russia 

But we can use math after the fact to figure out how much different stories mattered to the public and get some decent theories about how elections might turn out. One interesting thing is that the ups and downs of Trump's job approval haven't really been tied to Russia. They've been more tied to stories about competence -- when he lost the healthcare fight (while pushing unpopular legislation) or when he lost the shutdown fight, his approval dipped. 

It's impossible to predict everything about elections, but we can see real patterns. That's the stuff I'm most interested in 

Guessing Texas and California are the big prices, but what would you say for some of sleeper states on Super Tuesday or curious to how Democrats in those states will vote or caucus? Which of the non 4 early states would like to see more polling out of?

Great question. 

I'd be interested to see some more polling out of Florida, Illinois and Ohio. Those states vote on March 17, two weeks after Super Tuesday and each have a solid amount of delegates. They're all reasonably diverse states -- Democrats can use the results to gauge what some of the biggest pieces of their coalition are thinking. And if it's still not clear who the nominee is by that point, those states will likely have a real impact. 

Hi David -- thanks for this great chat and for taking questions today. This may be a hard question to answer, but what do you think "electability" really means? Is it a numerical term, or is it something broader? Up until the day of the election is 2016, I would say that a majority of voters really didn't believe Trump was "electable" -- and look what happened....the "numbers" only told a part of the story.

Electability is an extremely slippery term, which makes this both a tough and great question. 

When people talk about electability, they're usually asking about something like "wins above replacement" (I'm not a baseball expert so sorry if the comparison breaks down a bit). That is -- an electable candidate is someone who can get more votes than the average Democratic/Republican candidate. People can use their past election results, their appeal to key demographic positions, their ideological moderation or a whole number of other factors to argue that they're "electable" 

It's an interesting question on Trump -- I'd just point out that he was running against Hillary Clinton who was the second most unpopular presidential candidate in the history of polling (second only to Trump that is). If Trump was facing a different Democratic candidate, he may not have fared so well. And if Clinton had faced a different Republican, she may have not won the popular vote. It's hard to know these things for certain, and that makes electability a tricky concept. 

If polling turns out right and whoever wins Iowa, it will be a very narrow victory, does that mean that candidate only gets something like 2 more delegates that the runner-ups?

Yeah -- so the Democratic delegate rules are highly proportional. It's a little more complicated than this because some delegates are given out based on congressional district results, but essentially Democratic candidates who are above 15 percent divide the delegates up proportionally between each other. So you're totally right -- a close win doesn't actually translate into a big delegate advantage 

For what it's worth, I'm watching momentum and dropouts in Iowa. There aren't that many delegates up for grabs, but candidates who perform poorly often drop out. And candidates who do well can gain momentum and either seal things up or come out of nowhere. 

About the 4 non-early states, there has been a lot of polling out of Nevada, right? It is harder to poll because it's a caucus state? Even harder than Iowa with a smaller area and smaller population?

So in caucus states, polling is more expensive. Caucuses have lower turnouts than primaries, so you have to call a *lot* of people to get a big enough sample of people who actually seem like they're going to caucus. And Nevada is going to get less polling than Iowa because Iowa comes first, is flashier and generally attracts more pollsters. 

I don't know off the top of my head if Nevada is harder to poll than Iowa, but neither is a walk in the park from what I understand. 

Isn't Trump's whole MO to disrupt patterns?

This is a great question -- it is, and it isn't. 

Trump obviously breaks norms all the time. He acts in ways that we all *never* would have expected a president to act in, say, 2014. And he won the Republican Primary in part by taking on the GOP establishment and taking positions that were at odds with the consensus in the world of Republican intellectuals. 

But he's also been subject to some of the rules that constrained other presidents. The economy and President Obama's job performance suggested that 2016 would be a close race -- and it was. When Trump fails on policy, he loses supporters. Past presidents who had policy failures also lost followers. And as President, many of his major legislative pushes have resembled a version (maybe a worse version depending on your perspective) of what another GOP president might try. In some ways, he's a totally new president. In other ways, the laws of gravity still apply 

Did you see the story about the delegate stuff in Illinois' Democratic primary? The main thing I got from it was Pete Buttigeg's campaign (which has offices in Chicago) didn't get many delegates and did particularly badly in less white areas. Is the story being overblown or is it a bit of a canary in the coal mine for his campaign?

I think this is a real story. The Democratic rank-and-file is only 57 percent white, and anyone who wants to represent the party well needs the backing of the other 43 percent. The real test is mid-February though. If he performs well and Iowa and New Hampshire and still doesn't see gains in South Carolina, that is a big red flag for him. 

How much groundswell for abolishing the Electoral College do you think there might be in the US if the Electoral College denies the Presidency to the candidate who won the most votes? After all, in recent decades so few here have won the popular vote.

It's hard to say. It's possible people will agitate for the end of the Electoral College. Or they might just turn away from the process and get jaded (even more so than they are now). It's not a particularly fun possibility to contemplate, but still one worth thinking about 

Hi David! As a non-wonky, casual politics reader... I'm not totally sure how to know or guess what inputs to play around with. What are your guesses and inputs in the simulator? And what are you paying the closest attention to (fundraising, media coverage, etc.)?

Great question! I love that you're playing with the simulator. 

I think you could try a few different scenarios. Pick your favorite candidate, and give him/her a boost. Pick your least favorite candidate, and see what happens if you tank their fundraising. Maybe pick your favorite longshot and see how close you can get them to winning. I'd encourage you to make it your own and follow your curiosity. 

I don't know exactly what my guesses are -- I think the primary could turn out so many different ways. Usually, when I play with it, I make Biden or Sanders win because I think that'd have huge downstream consequences. I also do like to put Yang, Gabbard and other longshots in and see how high I can get them. It depends on how I'm feeling that day! 

I'm paying very close attention to polling. I'm interested in media attention -- who is surging, who isn't. I read a lot of the reporting about the ground game and who has volunteer power in Iowa. Fundraising is another important input. I try to track as much as I can really. But the most important thing, I think, is to track the polls and try to make sense of the narrative that's there. Because a lot of the other factors people talk about end up being reflected in the polls. Until Election Day, they're the best barometer we have 

I'm not very impressed with Iowans. They get A TON of money dumped into their state and a ton of pandering and that's what it feels like they are rewarding in the polling rather than some serious scrutiny of the candidates' record, proposals, etc...

You're not the only one who is unhappy with Iowa! There's a cottage industry of analysts who think long and hard about how to make the primaries better: rotating the order of the states, giving the party elites more sway (though voters might be pretty unhappy if that happened), clustering more states on the same day. A little searching and you might find a reform proposal you really like and can push for! 

Did Hillary Clinton "win" Nevada in 2008, but Obama got more delegates because he did really well in the caucuses in the more rural parts of the states?

I don't know about Nevada off the top of my head, but this is such an interesting aspect of 2008. Some of the reporting suggests that Obama had a more sophisticated approach to gaining delegates in a number of different states, and that helped him get his narrow edge over Clinton. I tend to think that's a move that only one candidate will ever pull off -- that is, every major candidate watched Clinton in 2008 and is probably taking efforts to make sure that doesn't happen to them. It's easier to game delegates in the GOP primary -- when the rules vary from state to state -- but it's still something to watch this year

Bernie Sanders refused to drop out in 2016 even when it became mathematically impossible for him win and Hillary also went to end when she wasn't going to win enough delegates so isn't it pretty much a given that all 50 states and D.C., Puerto Rico and the rest of the American territories will get contested primary?

I don't know that it's a given. In both of those primaries, Democrats had a clear winner by the end, and that may happen this year. If, say, Joe Biden and Bernie Sanders duke it out until June and Biden ends up with about 60 percent of the vote (one of MANY possible scenarios) he'd have the nomination wrapped up. 

The Democratic Primary process does tend to create long, bruising battles. Proportional rules make it hard to rack up a majority fast. And a contested convention is a real possibility. But we should wait and see how the early states shake out before getting too confident about that 

Got any thoughts on this? In December 2018 when it came out that he'd used his own foundation for personal gain, I was certain that this would cause his supporters to abandon him in droves. It was simple and easy to understand--the guy was stealing from his own supposed charity--and there was no question that it had happened. And yet, it wasn't even a blip on the radar. If the Senate convicted him, that'd probably lower his numbers a bit, but that hardly seems likely to happen. I await your response. Thanks!

My best guess is a recession. 

There's a divide within the Republican rank-and-file -- some voters approve of Trump "somewhat" and some approve "strongly." I think that if a recession happened, he'd lose the casual Republicans who don't love everything he does but are onboard because the economy is working well for them. 

Why did the President switch his official residency from New York to Florida?

That's a great question. I don't know, and there are a solid number of other questions I've got here. But this will probably keep bugging me for a few days until I look it up haha

Why not reduce the Electoral College's structural amplification of backwater voices -- by enlarging the House of Representatives, which would reduce the influence of rural votes at the Electoral College -- to address pressure to abolish the Electoral College?

There are a lot of potential benefits to increasing the size of the House (e.g. Representatives are closer to the people they represent). But I'd note that really the Electoral College favors big, close states. Candidates spend a lot of time in Florida not because it's rural but because they can get a lot of Electoral Votes from just tipping it a little. Pennsylvania has Pittsburgh and Philadelphia, and it still gets a lot of attention. The Electoral College arguably gives white non-college voters an edge, but the biases here are complex.

The only path I see for Klobuchar is if one (or two) of the top four candidates crash substantially. Do you see one (or two) of the top candidates crashing before the Iowa caucuses? If so, who? I recall in 2004 that Howard Dean and Dick Gephardt were leading in the polls a month before the Iowa caucuses and then crashed, leading to a win by John Kerry.

This is what makes primary prediction so hard. I could, in my mind, invent scenarios where any of the top four candidates crashes. Or where none of them do. In primaries, voters move easily between candidates and often don't decide until late in the process. So if you're a Klobuchar fan, I'd say you can definitely still hold out hope for a Biden or Buttigieg slip-up and for an opening in the Midwest/Mid-Atlantic relative moderate lane. But she's in the worst position of anyone in the top five 

So this is a fun thing to play with, although as you note it can't be good at predicting big surprises. If you build another version, I'd like to see you do one for a longer time frame and across states -- for example, if Klobuchar has a surge and finishes third in Iowa, how much does she gain going into NH and SC, and if she only holds in third there, what will happen on Super Tuesday? Or what would happen to Bernie if he finished third in Iowa, fourth in NH, fifth in SC?

Stay tuned ;)

I will definitely at least be writing about the sequential nature of primaries and how states influence each other. And there are a lot of possible additions to the simulator. Who knows what we'll throw in there by the end of the primary :) 

I wonder if, intentionally or otherwise, Michael Bloomberg's massive ad campaigns are really more of a Democratic anti-Trump campaign than him being a viable candidate for President. (Speaking as a Democrat, it's not as though I'd mind).

Maybe -- it's worth noting that part of the Bloomberg strategy is working though. He's made it to six percent in the national polls! But I do think that Bloomberg will have trouble when other candidates emerge from the early states. And for what it's worth, the effect of ads is typically temporary -- I'd guess that anything Bloomberg does now to hurt Trump will wear off by November 

I think the big example is Idaho and New Jersey. Obama and Clinton got about 50% each in New Jersey, but he got the same amount of delegates from New Jersey as he did from Idaho because he so dominated Idahoan caucuses.

Obama did get an advantage from caucus states in 2008. The number of caucuses is significantly down this year, so it's hard for others to pull off the same feat 

Do your models take into account efforts to suppress voter turnout through the various political machinations of states? Specifically, on the one hand Ohio and Georgia have had huge voter-roll purges. Also, how many Florida ex-convicts will be able to register to vote, or will Republicans find ways to obstruct them?

We think that a lot of that information gets rolled into the polls. Pollsters often ask voters if they're registered, how likely they are to vote, etc. and use all sorts of techniques to try to get a handle on what the electorate will look like. So we don't explicitly code that into a model, but some of that information makes it in by way of the polls.

And great question about Florida -- I've had my head down working on this model stuff, but that'd be a great thing to try to write about :) 

Does what he is doing wrong affect me or some real people is probably a good answer? His government shutdown effected real people. Didn't Kellyanne Conway say something to the effect that there is a difference between what effects people and what offend people?

I think that's a decent rule of thumb. I'd add that when Trump does something that a) effects people b) can be traced directly to him (e.g. a policy push he champions) and c) gets enough news coverage to break through the noise, that action might have an effect. And I do think his more offensive actions and words matter. They both helped him consolidate pieces of his base in the 2016 primary and alienate some voters who might have been onboard for a Marco Rubio or Scott Walker presidency. 

I am a big fan for many reasons (moderate woman candidate from the Midwest, for starters). What do you think her chances are?

Real but not super high. My gut answer is she's got about 1/20 or so odds. Her path is narrow. But, unlike at least a dozen or so other candidates who are running or have also ran, she has one 

This is the problem, because purged voters might not realize they've been purged.

This is a longer discussion, but there are pollsters who use voter lists and ways to improve on these estimates. I think this all makes polling more complicated and adds error, but it's not impossible 

Thanks for joining, everyone! Remember to play the simulator and subscribe to the newsletter. We have some great stuff in the works for the next few months. These were excellent questions and I'm sure we'll do another one before too long! 

In This Chat
David Byler
David Byler is a data analyst and political columnist focusing on elections, polling, demographics and statistics. He joined The Washington Post in 2019.
Recent Chats
  • Next: