Latest / Politics with Paul / How Predictable are Election Results? | Politics with Paul #117
Transcript
- Episode 117 How Predictable are Election Results: Welcome back to the Politics with Paul Podcast, your home for nonpartisan discussion of political science and civics. I am your host, Paul Rader. And today is going to be about how predictable election results are. Have you ever thought about how easily some of these elections are forecasted? There are a lot of things that you can already take as a given. Many contests are foregone conclusions for Democrats, while others are for Republicans. Certain demographics will sp clearly align with one major party or the other, and some individual voters make it painfully obvious who they will vote for. But while there is a lot about election results and voter behavior that you can see from a mile away, that predictability also depends somewhat on context. And it's those fluctuations where we see something unexpected that can make all the difference in who wins and who loses. And those unusual outcomes and data points, the upsets stand out even more because of how much we can predict already. And so today's episode of Politics with Paul is about which are the most predictable aspects of elections and which ones are harder to forecast or what can make them harder to forecast. And more importantly, we need to talk about why this predictability can be so problematic for democracy and and accountability from government and what we can do about it. So we first need to get into the difference between predicting individuals and predicting groups. Because there's a significant amount of misunderstanding about predictions and analysis. And it comes from this failure to properly distinguish individuals from groups. Pundits do it, politicians do it, campaigners, regular people, even academics can fall into this trap. And the quirkiness and surprises come from unusual occurrences at the individual level, much more so than they do the group. Although sometimes groups are fairly balanced towards one party or the other, but a lot of times they clearly have one lean one way or another. But you can get these exceptions to the rule that are what lead to these unusual outcomes and these upsets. But how do you know you are looking at an individual at one person versus a group? Or one example versus a an a group of examples. Because individuals are specific examples within a general group, as a rule of thumb. So some of these examples of the individual versus the group. When you're looking at voter demographics, groups generally go one way. towards one party or another, but the individual members might vote differently from the rest of their voting block. You're almost never going to get unless you really have a specific enough group where there's not that many people. You're almost always going to get an exception to the rule. No matter how much you think a group is going to be monolithic, there's always somebody that breaks that supposed rule. So you always got to be careful painting with a broad brush, not just because of potential. Statistical inaccuracy, but for the sake of not dehumanizing people and making them just seem like they should just be a statistic, disregarding their individuality. But when it comes to these voter demographics, the voting blocks tend to have a clear cut these ones vote Democratic, these ones vote Republican. For example, black women, as a general rule, overwhelmingly vote Democratic. But you're gonna find some black women individually who don't go that way. Or you look at white church-going Protestants, very heavily vote Republican, but you're gonna find some of those individuals who don't go that way. So if you look hard enough, you're going to find individuals bucking the trend. And enough if enough of them buck that trend, That's when you can find some of these really interesting results, these upsets in these elections. And so j groups are generally easy to predict which way they'll vote, or at least easier to predict. And to some extent how much they will. Like if you take those the example of black voters in general, they tend to, as a rule of thumb, and it might differ based on the election, but they vote somewhere Around maybe to 80 to 90 percent democratic. So if they start to go below that threshold where like maybe only 70, 75% of them are voting Democratic and the rest are voting Republican, whoever that Democratic candidate is in that election isn't they're in very serious trouble in terms of if they want to win. And so It's those individuals that make it tougher to gauge. Yes, you're talking about that general group of voters, but then when you start to pinpoint all the stragglers, so to speak, all those turncoats, as some people might say derisively, those are those individuals, and those are the ones that are tougher to gauge. Now, some of these individuals do make it painfully obvious, and we'll talk about how they do that. But you need to be careful with these broad generalizations depending on what you are analyzing or talking about. Even though a lot of times they're accurate. But you gotta make sure that you're talking about the group as a whole. You're not talking about individuals. if I'm gonna go up to one person and they, fit the stereotype I have in my head or that you have in your head of whether they should be Republican or Democrat or Independent, and you find out that they're not that way, well for one That's interesting. Wanna know why exactly they think differently, but also you can really offend somebody there, or you can make them think, Well, how could you possibly think I don't have my own viewpoint about things? But some individuals are predictable. And these parties and campaigns have a staggering amount of info on you that you either have no idea they have, or you may not think that they would predict your voting behavior. Some demographic info you might expect that they can use to predict, but businesses that you frequent, even bumper stickers on your car can be used to gauge your political views and the likelihood of. who you'll vote for and with sometimes disturbing accuracy. Now, some of those things that you use, they I they tend to give a sense of who who your group affiliations with are with, and so that will be used because we have an idea of how these groups generally vote, that might be what is used to indicate who you vote for, but a lot of times we don't stray from that social group and how they see things. But you still sometimes get People who think differently. But apart from that, what else is used to predict elections? You have a bunch of these election forecasters that exist to gauge how likely it is a party or a candidate are going to win which elections. And some of the biggest ones, and we've used them several times on this show Cook Political Report, Inside Elections, Sabato's Crystal Balls, CNalysis. Split ticket, elections daily, there's a lot of them, but those are some of the really big name ones. And all of these forecasters have their own formula that's either proprietary or at least they keep it secret, the exact formulas. And they use these to determine how safe or competitive a seat is for a party or a candidate. So if it's almost definitely going to go to One particular party, it might say it's safe Democrat or solid Republican. And if it's competitive, it'll say like it's a lean Republican, it's a tilt Democrat, or it's a toss-up. And so while we don't know the full exact calculations that all these groups use, we can still point to some of the most common criteria they use because they still often reference some of the data that they use in their write ups and say, Why did we change this election rating from lean Democrat to toss up or for example? Well, it's because of X, Y, and Z. And some but not all of those most common data points that they use are polling, voter registration by party affiliation, if the state tracks that, the history of the district or the state in terms of its election history. And demographic trends of that district. So when it comes to polling, and for bigger races that get more attention, the trend of a collection of polls, that is a very common thing to use, and it's it a lot of times it's polling has faced a lot of struggles in recent years. Sometimes those are overblown, though, because you have a lot of people that think, polls suck and they're They're completely wrong. They're they're they're they're inherently not trustworthy, all these things. I've I've talked about it on this podcast several times. that that is not the way to treat polling. They are generally accurate. The ones that are wrong just get them most scrutiny because people the only thing that people find more interesting than polls are polls that are wrong. And media also problematically just run with any polls that seem interesting to them that they can use to tell a story. And sometimes that propagates fake polls, ones that aren't that are purposely made to lead people astray, or low quality polls. But the vast majority of these polls you see are not meant to be that. And sometimes these polls are for that specific race, like a US Senate race in whatever state is expected to be competitive. And sometimes they use what is called generic congressional ballot, which is just simply it doesn't give you specific names of people. It just says all else equal, would you vote for a Democrat or a Republican for the this for Congress? And then you say Republican or Democrat. And so if you have a large lead in a poll, as long as it's a reputable pollster, got gotta make sure that it's not just any rinky dink poll, it's going to suggest that that election is more predictable. Because if they've done the proper sampling and they've weighted it and done all this complicated statistical analysis on it and put it out there and it's a reputable pollster, there's a very good chance that that. poll, the the results of that poll are going to give a good indication of where that race stands, especially as you get closer to the election. Partisan data of voter registration, if that state tracks it. So as as of August twenty twenty six, thirty one states track registered voters by party. Party affiliation is the number one indicator of how people are going to vote. And there is more straight ticket voting than ever before. And so if a district or a state is lopsided towards one party, that district or state is going to be much more predictable in terms of who they're going to vote for. If you look at statewide for Wyoming, it is overwhelmingly Republican. They are if if you're running statewide, it is almost guaranteed that you're gonna win if you're a Republican. In a lot of areas of Wyoming, you're very likely to win if you're a Republican. Every s there's a few pockets that will be democratic, but generally speaking, you're a Republican. And likewise, Maryland is quite decidedly Democratic. And so if you're running statewide in Maryland, it's incredibly likely that you're gonna win if you're a Democrat. But you'll still have those Republican pockets. So it it it remember the district is not the same as the state. It's almost like that individual and group dynamic that we talked about before, where each of those districts in a sense can be an individual part of that state, and the state being the whole group. But some of those districts are very lopsided. A lot of those districts are very lopsided. The election history of the of that district or state, so not just what it is right now, but how it's voted in the past. If that area has a habit of voting for one party, especially if it's been fairly handily, there is a good chance it will keep voting for that party that way. And so that if if There's a history of voting for one party, it's more predictable as to who's going to win that that district in the future. But if that area tends to have close elections, it's probably gonna have close elections again, and who is going to win will be less predictable, although you can fairly you can reasonably predict that it's gonna ha be a close election again, w whoever wins. But you still have to look at the trend of the margins of victory. So if six years ago the margin of victory was fifteen points, where whoever won had fifteen percent more. So like if it if the winner had fifty percent of the vote, and then the second place had thirty five percent of the vote, that's a fifteen point margin of victory. But then four years ago it was a ten point margin of victory, and then two years ago it was four point margin of victory. The history of that election suggests that now this time, okay, maybe this is actually gonna be a much more competitive race. Maybe it's even going to flip because although one party has been winning frequently in that district, it's become a lot tighter over time. And on a related note, you also look at the demographic trends of that district or state. Of that district or state. Most demographic groups, like we said before, have a lean towards one party or the other. And if more groups leaning towards one party start to make up a larger portion of that electorate, then that district or state will start to become more predictable. And even if the state doesn't track the party affiliation of those candidates. or excuse me, of those voters, they may track other demographics of the voters that give a sense of the party makeup of that state because those demographics tend to fall one way or the other and can use that based on public opinion and national data to get a sense of where those voters lie. But besides just simple well who who is this particular contest going to be won by a Republican or a Democrat, there are other aspects of elections that we can say with fair certainty are going to happen. And so let's look at a few of those and see what exactly is predictable about them. Something that is just about guaranteed to happen, at least t to s to varying degrees depending on what type of office you're looking at and at what level. Incumbency win rates. In Congress, those who already occupy the seat that they are running for re-election in will win upwards of 80 to 85% every election year. And a lot of times that's lowballing it. They win oftentimes even more than 90% of the time when they're running for re-election, which is an illustration of Fennow's paradox, which is basically that Well, everybody hates Congress. Congress gets all these incredibly low rates of approval, but yet Congress gets reelected at sky high rates. Because people tend to either love their congressperson or they're just not sure enough about their congressperson's challenger. They know who their congressperson is. And they're not even if they're not super enthused about their congressperson, they're not necessarily angry enough or Disenchanted enough to vote for somebody else. And while stats aren't as readily available for incumbency win rates for state legislatures, despite how many seats there are there, the vast majority of those incumbents will win as well. Most types of elections you look at, incumbents are going to win at very high rates. very predictable aspect about politics is the number of partisan wins, whether d a Democrat or a Republican wins versus whether an independent wins, third whether that's a third party or a no party candidate. The overwhelming majority of partisan elections are won by either the Democratic or the Republican Party. And third parties and independents win only a mere fraction of those races. And this is not counting nonpartisan elections because those are by default nobody can run as a party. Although people who would have run as a Republican or Democrat in those races, they still tend to have more money. They still tend to have more connections. And so they still tend to win those anyways, even if they're nominally not running as a Republican or Democrat. So it is very difficult for independents to win. And there are many reasons for this divide, despite how many people identify as independents. And I talk about a lot of these things In my book, Why Independents Rarely Win Elections and How They Could Become More Competitive. A lot of reasons why independents win at such low rates is there's the fear of the spoiler effect. There's ballot access that prevents a lot of independents from even getting on the ballot, let alone being voted on. Independents having far less money than the parties do typically, and they have fewer connections. That are already in office because Republicans and Democrats can often rely on other incumbents to at least give them not necessarily monetary support, but they can bring their own networks in to try to back these other Republicans and Democrats. there's less room for error in media coverage for these independents. As soon as an independent makes one flub. Everybody takes them far less seriously, and it hurts them a lot more than it does a Democrat or a Republican, typically speaking. And they also tend to have less r name recognition. Now, independents do have a much better chance when either the Democrat, either the Democratic or the Republican Party don't run a candidate in an election. So if it's just a Republican versus an independent or a Democrat versus an independent, but even then they still often struggle to win elections. Another predictable aspect about politics is the frequency of some voters to turn out. Although there are some fluctuations here and there, we generally have a good sense of what turnout rates are going to be. Presidential elections always get the highest turnout rates, and then that's followed by midterm elections. And these voters who have come out for midterms in the past, they're probably going to come out again this time. But if it's voters who only come out for presidential elections, They're s very unlikely to come out for a midterm because presidential elections are the crown jeweled. They're the ones that get the most fanfare. Midterms get a lot of fanfare, just not as much as the US president. And a lot of people think that the US presidency is the only thing that matters. Or they just don't want to take the time to research all this other stuff. So they just don't come out for midterms. They let other people decide that. And also non voters tend to stay as nonvoters. If they haven't vo if they've had plenty of opportunities to vote in the past and they haven't taken them, they're probably not gonna take them now. But every so often you get individuals who do break the norm. Donald Trump was able to do that for many voters in twenty sixteen. Barack Obama was able to do that for a lot of voters back in 2008. There's also another aspect of politics that it and election results that is very predictable is you'll see ballot drop off. As you further go go de as you go further down the ballot, the more you will see voters skip those contests because ballots tend to have a lot of things to vote on and sometimes people get tired of voting on things or they just only want to research the highest things and they weren't confident about what else was on the ballot. And so you often see local races get the fewest votes because they're the ones that are furthest down the ballot. And so It's a it doesn't really matter what election it's in. Generally speaking, the further down the ballot you go, whatever year it is, you're gonna get less participation. And then lastly, midterm elections you can predict with fair certainty are not gonna work out for the president's the sitting president's party, because they're almost always bad news for the sitting president's party. You can just go back, especially over the last ninety years. Of midterm elections. At least for the US House. The US Senate is murkier because only a third of those seats are up. And it depends on how many for each party are up. So if a Republican is the sitting president, but more seats that are up for reelection in US Senate are Democratic. Things might not work out so poorly for that sitting president's party in the US Senate. But the US House almost always a loss of seats, and a lot of times it's a staggering number of seats. Barack Obama in his first midterm election back in 2010 lost s sixty plus seats. Democrats lost sixty plus seats in Donald Trump's first term, midterm back in twenty eighteen. His party lost over forty seats. And it's not necessarily the the the number of seats lost isn't necessarily that high, but you still tend to see them lose a lot of seats. And because Republicans right now have such a razor thin margin of b majority If Republicans even lose five to ten seats, it's going to completely flip control of the of the US House. And so the n the exact number of seats, won or lost by a party, is not necessarily that easy to predict, but a certain range is dual based on the calculations of polling and demographic trends and all these other things that we just mentioned earlier. But what else can impact The predictability of elections. You have other contextual data that can make election results either more obvious or more unpredictable. And for those I want to highlight here, the national political climate. Politics has become increasingly nationalized, and national politics, particularly whatever happens with the president, has a trickle-down effect. So when, for example, you take Presidential approval rating. Not the end all be-all, but it does tend to have an effect on the president's party. So if the president is not very much approved of and an election is coming up, that president's party is gonna feel the heat. And if that president is more approved of, then his party is probably gonna do a little bit better in that election. But this becomes especially pronounced in midterms when the president is not on the ballot. But many people turn out against the president's party and endorsed candidates. Other races on the ballot can also make for a more predictable or more unpredictable election. And this can go both ways. Either a higher level race can affect a lower level race, or a lower level race can affect a higher level race. Oftentimes it's ones that are higher on the ballot impacting a lower one. So if there's a US Senate race on the ballot at the same time as the US House for your district, well, or your state, well, that US Senate race can have that trickle down effect on the US House. Or even on state legislature. But sometimes you'll see a lower level contest make an impact on a higher level one. So if there's a salient ballot measure for you to vote on, it might make a difference in a tight US House or governor's race. Different states political contexts are another thing that can affect predictability. Because some states are just wackier than others and are inherently more unpredictable. A couple of examples I use for weird politics are Alaska and Maine. And that's not anything to do with whether I agree with them or not. They just have weird things about them that you typically don't see in other states, especially at the state level. With Alaska, you have a top four primary that leads to their general election having four Maine candidates. And they also In terms of voter registration, have tons of both non party and undeclared voters. So difference there being non party means you specifically, I forget exactly what they call I think they call them NPAs. No party affiliates. They specifically said they don't identify with a party. Undeclared just register to vote and they don't say anything about where they lie. They don't even say that they're not affiliating with any party. They just are registered to vote and that's it. Alaska's state legislative chambers are often also controlled by cross-party coalitions, which is you almost never see that kind of thing. Maine has an unusual number of third party and independent state legislators. Now, still not that many, as they're still dwarfed by the number of Republicans and Democrats in that legislature, but Compared to other state legislatures, you do tend to see more third party and independent legislators in Maine. And they also have an independent U.S. Senator, Angus King, although he still caucuses with Democrats. So some states, because they're just wacky by nature, compared to other states' politics, sometimes they make things more unpredictable. And then lastly. Candidate strength. Because for all the power that the party affiliation has, where some people will just vote Republican no matter what, some people will vote Democrat no matter what. And these and they'll just, regardless of who their party puts up, they'll look past those flaws. Having a particularly weak candidate can tank that party's hopes in that election, no matter what other advantages they may have. And they may have a lot of money, they may have a lot of connections in government already that are trying to help them, but it still may not matter. The candidate's personality, charisma does have a may big impact on your ability to win an election. Doesn't mean that charisma will win you the race, but it definitely helps. Scandals and gaffes, any foolish things you say, any criminal or at least unsavory things that you've been a part of, those can tank your your ability to win. Your ability to fundraise and gain ador endorsements or lack thereof and more of these things affect the strength of that candidate's campaign and therefore the ability to win that election. So while a lot of races will you could say with almost certain complete certainty that that party is going to win that election If that candidate screws up enough, that might not actually be the case. So that is all that I wanted to talk about when it comes to what makes these races more or less predictable. But a more important discussion point is how predictability can be problematic. We've established how there can be a lot of foreseeable outcomes in elections. and data points and voter behavior and so on. But what are those implications? many things, there needs to be a balance of some sort. And unfortunately we lean too much into being predictable. And that's not a new phenomenon. This isn't just like, we were never able to predict how people were able to vote before. But with more historical data at our fingertips, the means to get this information Increasingly predictive technology, especially with the rise of artificial intelligence, power being concentrated even more within the hands of certain individuals and groups of people, that predictability is being harnessed arguably more effectively than ever before. And usually it's not for a good purpose. But why would it be a problem, this predictability being harnessed more effectively? Because it's another tool that could be used against you by people who are more interested in power and prestige than trying to implement good policy or effective governance. Because they can manipulate your biases in their favor. They know we have inclinations towards one party or another, or we care a great deal about several specific issues, or maybe One particular issue. They see even the kinds of places that we frequent for business and how much that can indicate how we vote. They can manipulate all these biases in their favor, not just of who you'll vote for, but how to make your loyalty staunch. So they can either give you slanted information, bias information, or they can flat out give you f False information or information that's even slightly faulty to try to mislead you or to try to control effectively control how you vote and who you vote for. Political interests know that they have bases in the electorate that they can Always rely on no matter what. And it keeps them in power. And a lot of times we think it's not a problem when they're interests that we tend to align with. Because it's that in-group-out-group bias where, well, if they're part of my group, they're they tend to be good guys, even if that's not even close to the truth. But not only that, if you value the privacy of your vote, that privacy is. continues to be diminished because of all the data that's available to predict your vote and the data that can be used to manipulate your vote. So it's always good to reevaluate your biases so they can't be used against you. We're never going to be 100% unpredictable and we don't need to strive for that, whether it is as a group or as individuals. We don't need to be completely unpredictable. In fact some predictability is good for when we build relationships with each other and being able to communicate with each other. If we couldn't get any sense of how other people thought or acted, it's really hard to connect with those people. Especially when you have certain common interests, you're going to have some sense of how and certain shared personality traits. You're gonna have some level of understanding of who that person is and being able to connect with them. So you need some level of predictability. But when it comes to our voting choices, we become too predictable. And it's being used against us. And we need to be aware of how much that predictability is employed against us and push back against it. And that starts with being more aware of that fact and deliberately trying to make ourselves less Predictable. And a lot of that also comes from breaking out of our echo chambers, out of our silos, and deliberately trying to engage with people who don't think the same way that we do. So for the election tomor tomorrow, well, when this episode comes out, it will be the day of the election. And for elections of the future, always keep this in mind and try to recognize there there are gonna be some things that are just they're gonna happen no matter what, but we should also try as much as possible to be less predictable. Because otherwise we can't hold government accountable because they can just use all these predictable aspects against us. And so that's where I'm gonna end today's episode of Politics with Paul. As always, thank you so much for listening. Please like, comment, share, subscribe to the podcast. My links in the show notes, like my five-day educational email course about how we can all more effectively mitigate toxicity and polarization and the ways that we can fixing the ways that we accidentally make it worse, even when we're trying to make it better, how we can actually turn those things around. My book on political independence that I referenced earlier, and my writings on medium dot com and then my other links like my LinkedIn, my Facebook, my website, and my email, Paul Raderwrights at gmail.com. So thank you again, as always, and we'll see you next time.