Trump’s Coalition

For all of the time we spend looking at electoral maps to predict presidential elections, we hold a lot of conflicting ideas about how regional variations in demographics affect voting behavior. In the Jim Crow South, V.O. Key (who had pretty good maps for 1949) notes the relationship between the percent of the black population and support for conservative Democrats in the south (African Americans could not vote in the south). Southern Republicans at this time tended to live in Appalachia, some of the poorest and whitest parts of the south. Eastern Tennessee, Southwest Virginia, Arkansas, and parts of Northern Alabama served as the stronghold for the more liberal Southern Republican Party at this time. The regional variation in racial composition (which also correlates with arable land) was a strong predictor of voting behavior. In much of the South still today, whites, in counties with large black populations are typically the most conservative voters, and those most loyal to the Republican Party. This has always been thought of as a purely southern phenomenon but I wanted to see if there was any evidence of this in the much whiter midwest where Trump fared much worse than he did in the South or Northeast.

One of my goals at the end of the primary process is to create Republican and Democratic maps of all of the primary and caucus results to highlight the regional patterns from this election. Real questions about the role of class and race on political attitudes have correctly renewed attention to regional variation in the United States. Since V.O. Key we’ve know the role that region plays on voting behavior, and this primary season likely offers a lot of insight into those same questions. Below I’ve made an interactive map of the Midwest, excluding the caucuses of Minnesota and Iowa, which tend to be unrepresentative. If you want to fight me over my definition of the midwest, go ahead, but the map below shows the core, indisputable states of the Midwest. As you’ll note from the map below, there isn’t much of a relationship here.

Midwest Interactive Map: 

Indiana is an outlier because in someways it is a southern state trapped in Midwest. The only reliable GOP state in the midwest, it’s also the birthplace of the KKK. Trump does better than you’d expect in Indiana, but it seems like by the time we got to Indiana, Republican voters had already coalesced around him.

However, there are regional models of voting behavior that instead focus on class instead of race. In Andrew Gelman’s Red State, Blue State, Rich State, Poor State, he notes that in America, there’s a good relationship between income and vote choice. Higher income Americans vote more Republican, lower income Americans more Democratic. However, when you divide states up into Red and Blue states, the relationship changes. In rich states, which tend to be blue, social issues predict voting behavior, and income has a weak relationship to vote choice. In poor states, which tend to be red, income explains voting behavior much more strongly, with rich voters in poor states being very conservative and Republican, much more like the national pattern. Thus Gelman asks, “What’s the Matter with Connecticut?”

On The Weeds podcast last week, Yglesias and Klein do a great job breaking down Trump’s rise, particularly his overt and covert attacks on multiculturalism and the changing economic forces in America. Yglesias speaks at length about how in 2008, he thought the common theme that kept the GOP coalition together was its militarism toward the Middle East and it’s growing xenophobia/homophobia. Trump reanimates this ideological force, and his support for Social Security and Medicare falls right in line with many of these voters. Barack Obama’s presidency has brought multiculturalism to the forefront of the Democratic Party, pushing class issues toward the backburner, much to Sanders’ supporters demise. Cultural changes, even those that have nothing to do with race like the internet, have people on edge. There are ironically enough viral FB videos about our use of phones that you could watch a new one everyday for the rest of your life. Trump has staked his claim to “Making America 1985 Again!” while allowing different groups of working and middle class Americans, who don’t see bright futures for themselves, to come into the fold. His support of the existing welfare state, instead of turning the safety net into torn hammock like his Republican rivals has helped him with a certain kind of independent, those who would most likely be Republicans, but who don’t often vote and have weak non-ideological allegiances.

One of the ways you can tell that Trump’s appeal is not ideological is that he does well everywhere, but each for different reasons. Trump does well in the South despite a documented record of not caring at all about social issues, or Christianity in general. His xenophobia, misogyny and racism, plays well in the South, but it does not explain his dominance of the northeast. In the Northeast it seems that Trump’s focus on manufacturing and aggression on ISIS has helped with rural and working-class suburban whites. Trump winning Arizona, or South Carolina makes perfect sense to me. But New York, Florida, Michigan, and West Virginia are much harder to understand until you realize that Trump’s campaign is based upon opposition to the changing forces of economic and social life in America, not ideological positions, or party orthodoxy. Republican elites, co-opted by a libertarian approach to government in recent years do not offer anything to huge swaths of the public whose quality of life is declining in an international marketplace. Trump is running a reactionary campaign that longs for things to be like they once were. It’s not 1985 anymore, and that’s a good thing, but a lot of Americans (not even a plurality), wish it were.

Off Topic: The Intersection of Reading and Data

I’ve been fortunate enough recently to be able to take an extended leave of absence that’s allowed me to pour over electoral data, as well as actually keep up with my extended reading list. It’s the closest thing to a sabbatical I’ll likely experience and it’s been mostly great. Having the opportunity to actually sit and watch the news cycle in real time I’m increasingly convinced that much of Twitter is repetitive noise, with a few good articles waiting to be mined. I don’t have any interest in blasting journalists, but an intransigent Congress has made most reporting speculation about electoral tactics and messaging. Coverage is almost entirely process, because there’s no substance to cover, not because journalists are evil. Twitter just amplifies this coverage that I think makes it seem like political coverage is worse than it is.

I’ve always held that reading whether academic work, traditional journalism, or blogs like this one, is both necessary to be a useful member of society, and to also grow as a person. I don’t think that people become conservative over time as a function of the aging process. Rather, the fact that people stop encountering new ideas and stop having their beliefs challenged, makes them closed off from political movements and social undercurrents, they’d otherwise have seen. With the exception of Transparent, the overwhelming majority of support for Trans rights has come from academic circles publishing in Slate, Tumblr, and progressive publications like The American Prospect, or Mother Jones. It’s a movement that outside of parades and protests in major cities, likely goes unnoticed in suburban America. If you’re not actively keeping up with one of several publications you’re missing a trove of information about the evolution of social norms. As a devout Clinton supporter, there’s quite a lot that could be improved in her rhetoric and policy prescriptions if a top Clinton aid were skimming through Jacobin, a smart quarterly, with their fingers on the pulse of young ‘middle class’ America. I’ll freely admit that other than David Frum’s Twitter account and the occasional post from Daniel Larson at The American Conservative, I’m out of step with what’s left of the intellectual conservative movement.

This form of writing has yet to really engage with the growing empiricism and ‘data science’ you see at FiveThirtyEight, or the really great collection of sports analytics writing like Chris B Brown, Kirk Goldsberry, and Chase Stuart. Nate Cohn in my view has easily been the best at contextualizing electoral data from the primaries to really explain the dynamics of the 2016 race. Harry Enten has really done a remarkable job tracking polls and helping you see through the smokescreen of erratic pollsters like ARG that have massively biased polling averages. Writers like Jamelle Bouie are great at using social science literature and empirical data to map out the electorate particularly the Democratic coalition that isn’t fraying, but is certainly starting to diverge. As the Democratic Party seeks to capture the American center, its leftwing is firmly standing its ground. There’s an enormous amount of political writing, and a lot of it is actually really good, but it’s only getting to people with an unrealistic and unsustainable drive to sift through the noise.

If you’re not willing to engage with academic literature (a sentiment with which I’m very sympathetic), to grasp American politics requires both a passion for American democracy and a statistical expertise that likely has to be taught in school, but that undergraduate courses largely fail to support. I’m currently spending countless hours everyday following the news, and I feel that the majority of what I encounter doesn’t add anything to what was written the day before. The more time I spend reading the news, the less satiated I am by it.

Part of my frustration though is that it is extraordinarily difficult to both write well while still presenting and describing even simple empirical data. To be politically literate requires you to understand both, while very few people have the capacity to do it. We need more academics who can actually write to broader audiences, but the kind of person who writes a 400 page dissertation is not likely the best at condensing information into a digestible format. We’re a long way from sorting this out.

Racial Homogeneity and Social Welfare

It is well documented in the comparative politics literature that the racial homogeneity of a population correlates with greater social welfare expenditure. The logic is that people are more willing to accept taxation for redistribution when they believe that the money is going to help people like them. This is one of the reasons why people like David Frum fear immigration so much, in that increased welfare expenditures on immigrants very well might decrease support for redistribution in general. Zizek himself has similar thoughts as well. The two figures below, though old, show this relationship pretty clearly.




However, I’m not interested in this relationship abstractly. I’m particularly curious how it pertains to Bernie Sanders voters and future support for social welfare in America. I need to break down this argument carefully, so I’ll start with a macro-level approach to American elections first.

Since the 1960s it has seemingly been the case that younger voters, particularly college students hold more liberal views on social life and economic policy than the rest of the country. But what happens to those students’ political views when they leave college campuses? Political science can tell us almost nothing about the way political attitudes change over time, because we do not have panel data that tracks political issues over more than a 2-4 year period. We have reason to believe that people get slightly more conservative over time, but how that happens and why remains unclear. For a candidate like Bernie Sanders to win in the future, they will need Bernie’s current supporters to stay as economically liberal as they age while building a larger coalition of liberal activists. An assumption here is that a Bernie-style candidate will continue to consolidate support with the groups they already do well with and expand its base. This is an end of history argument that you see a lot when writers only have to produce copy for next week, with almost no consequence if it ends up being totally wrong. Assuming that this coalition will stick together past 2018 is hard for me to believe.

This is the story told by really smart writers like Greg Sargent and Matt Yglesias who I respect and read regularly. But I’m still skeptical that the future of the Democratic Party is with a socialist candidate. This is not meant to be an exercise in hippie-punching, but a sincere concern that social democracy will not spread outside of the largely white, liberal enclaves of the Pacific Northwest and New England. Much like Hillary’s wins in the deep South, Bernie’s wins in the Non-Colorado Rockies and plains-states are important for the nomination, but not the future electoral stakes of the party. If in fact social welfare support is dependent on racial segregation or at the very least highly correlated with it, it is increasingly unlikely that there is a winning electoral map for that style candidate in the future.

One of the more fascinating electoral changes under President Obama has been the polarization of the white vote. In 2012 Obama got 39% of the white vote, which is much worse than the losing efforts of Gore and Kerry. In the 2016 Primaries Alabama serves as a great proxy for this change. In 2008, white voters made up 44% of Democratic primary voters, and black voters 51%. In 2016, black voters made up 54% of Democratic Primary voters and white voters only 40%. The pattern was even more extreme in South Carolina. In 2008 white voters made up 43% of voters in South Carolina, but only 35% in 2016. Black voters increased their share from 55%-61%. As the black share of the electorate increases, the white share of the Democratic party declines.

A real problem for a candidate like Bernie Sanders is that demographically similar white voters in Washington state to those in Ohio or Florida respond very differently to Bernie’s rhetoric, much to his dismay. The biggest impediment for Bernie Sanders is that the states that really matter for winning the presidency are Ohio, Florida, Virginia, and Colorado. These are states where the white voters who make up both the Democratic Party and the ‘winnable’ moderates are the people who Sanders has not made strong enough inroads with, and who he may not ever be able to reach because of latent racial animosity of whites. Working class whites and liberals in the largely white communities of Madison Wisconsin, Seattle Washington, and Manchester New Hampshire will continue to support candidates like Senator Sanders. How do you get whites in Raleigh North Carolina, Cleveland Ohio, and Chicago Illinois to follow suit? What I want to make clear here is yes, Bernie wins about half of white Democrats in these places, but whites in racially diverse places are much more conservative than their incomes would predict, meaning there are just fewer whites here for Bernie to potentially win. As the country becomes more racially diverse I worry that white voters will become increasingly conservative and support for his style of social democracy will fade with it.

This raises a question about what it is that’s driving people to Bernie Sanders. If you read Jacobin, it’s largely a class based appeal that Bernie Sanders is going to fight income inequality, and no longer accept the influence of the financial industry in positions of government or in fundraising events. Younger voters, saddled with student loan debt, poor job prospects, and the fact that having a child will be increasingly difficult has helped him  solidify this message with a supportive base. But Bernie has both an economic and an identity politics appeal that American elites, even those in the Democratic Party, are both greedy and corrupt that can no longer be allowed to ignore the interests of the American people. Though I disagree with Bernie’s diagnosis that reforming campaign finance would allow liberal initiatives to flourish in Congress, his working-class focus against elites is one I can imagine being successful given progressive traditions in american politics. That “the economy is rigged against working people” is an appeal that will work as income inequality grows, which it likely will even with large tax transfers, and one that will likely appeal to African Americans as its focus extends beyond the economy to more explicitly focus on racial and gender inequality.

If class is the dividing line for the American electorate, then we’re not far removed from a Bernie Sanders-style presidency. But if race is the dividing line, the future of the country and the Democratic Party is not so clear.



Quasi-Super Tuesday Dem Predictions

The Midwestern Democratic races are going to be extremely close, with Hillary holding narrow leads in Ohio and Illinois. The campaigns are tied in Missouri, though with only two polls so far it’s hard to make heads or tails of it. I have a feeling that much like last week, Hillary and Bernie will win about the same number of states, while Hillary expands her delegate lead.

Click her for an expanded view of the figure below. SuperTuesdaydem15


In fact I expect Bernie to win all three of the midwestern states, while Hillary increases her delegate total. Bernie will fundraise massively off his wins, the story will surround Bernie’s surge, but there just doesn’t seem to be a way forward for Senator Sanders to win a majority of delegates with the remaining states. Hillary is leading every poll in Ohio and some operatives have expressed optimism for her there but I think the momentum is with Sanders. If Hillary were to win either Illinois or Ohio I think it would seriously quiet Bernie’s surge. Winning 3 or more states and a lot of delegates would probably end the viability of the Sanders campaign.


Super Tuesday March 15 GOP Polls

Below are the GOP polls for tomorrow’s contests. Florida and North Carolina look like locks for Trump, and given his dominance of the South excluding Cruz’s win in Texas these states will likely follow suit. I’m less confident in Ohio, where Trump is losing and Illinois and Missouri, where there’s less polling and is the part of the country where Cruz has had more success, given his wins in Iowa, Oklahoma, and Kansas. Cruz has a real operation in these states, and it wouldn’t shock me if he took both Missouri and Illinois, though I’d wager Trump wins at least one of them.

Click this link for the expanded view of the figure below. SuperTuesday15


I think there are only two public polls in Missouri, and I could only track down one of them. Trump is winning, but this poll only had 208 respondents. I think Missouri is competitive, and I think Cruz is more likely to win Missouri than Illinois.


In terms of Delegate math Trump should take all 99 of Florida’s delegates and about half of North Carolina’s 72. Kasich should win Ohio’s 66 since Ohio like Florida is winner take all. Illinois is winner take most, I think Trump is the favorite so we should expect him to take about 75% of the delegates there, though if he loses to Cruz, he’ll only bring in about 20% of Missouri’s 52 delegates. Cruz will likely win the Northern Marianas Islands given his caucus success, netting him ~5 of the 9 delegates. By this back of the envelope math Trump should win around 160 delegates. With Cruz falling in the 60 delegate range unless he steals Illinois. Kasich, even with an Ohio win will likely only get about 70 delegates which helps him play spoiler but keeps him in 4th place. If Cruz steal Illinois and Missouri and Kasich takes Ohio, it becomes much more challenging for Trump to get to 1237, though he’ll still have a large delegate lead heading into the convention. Chaos awaits.

Michigan Maps Interactive

In my spare time I’ve been trying to learn how to make spiffy maps of electoral results like the ones you see on the New York Times. The two maps below are attempts at doing just that. The first map is of the Democratic race and shows the Hillary and Bernie vote shares respectively (I omitted votes for other Democratic candidates for simplicity). Hillary simply did not run up the score in Wayne County like she should have. She needed 70% of the vote in Wayne, but didn’t get there as the first map shows. She didn’t get to 70% anywhere in Michigan, whereas she got more than 70% virtually everywhere in Mississippi.


Interactive Map: Enjoy

Click on the link below to open up the Interactive map in a new tab.

Michigan Democratic Map: 

Trump dominated in Michigan, getting ~37% and doing well throughout the state. Trump won the eastern and northern portions of the state while Cruz’s support consolidated in the West. Kasich did well in the Southern portion of the state, namely the Detroit Burbs and Ann Arbor. Rubio had a downright awful night everywhere in the state.


Michigan Maps With Layers:

Click on the link below to open up the Interactive map in a new tab.

Michigan GOP Map:

What Part of the Party Should Decide?

As Hillary slogs to the Democratic nomination, a criticism of her path to victory concerns the role that Republican states, namely the South, have played in accumulating her large delegate lead (excluding her massive superdelegate count). This critique is most vocal from Bernie supporters who are frustrated with Hillary’s huge margins with black voters, particularly in the South. Though in the most recent YouGov poll in Michigan, Hillary has a 61 point lead with black voters there as well. The frustration from Bernie supporters stems from the fact that their policy agenda hasn’t caught on outside of young Democrats and ‘very liberal’ identifiers.  Michigan_Dems

But this conversation does bring up an interesting point. Who should decide who the nominee is? By this I mean should party nominees be decided by base voters in solidly red or blue states with closed primaries, by swing states with open primaries, or by the process we currently have with a little bit of everything? A related question: What are primaries for? Should the nominee be the candidate that best represents the base, or should it be a process for choosing the candidate most likely to win in November? I personally subscribe to the latter view but it’s worth discussing. The GOP, which controls the legislatures of virtually every southern state moved up their respective primaries to the SEC primary (Super Tuesday) to ensure that a conservative base-approved nominee would win the nomination. I don’t think they were hoping for a Cruz-Trump sweep of the former Confederate states, but that’s what they got. I think it’s fair to say this process ensured that the Republican base has more say in selecting the nominee than the GOP leadership does. What you think about this relationship probably influences how the nomination process should play out.

The Republican base, which is more conservative than the Democratic base is liberal, seems disinterested in electability as a criteria for choosing a nominee. The Democratic Party has a slightly different dilemma. On paper, and based on every understanding of the median voter theorem, Bernie is less electable than Hillary. And even within the Democratic party, Bernie is beloved by younger Dems as well as the more economically liberal parts  of the base, but struggles with moderate and older Dems. To what extent should ideological parties appease their most ideological voters?

In the Democratic Party though there’s another tension. Without black voters, Democrats could not get to 270 electoral votes as black voters are crucial to winning Ohio, Florida, Virginia, Nevada, Pennsylvania, Michigan and in competitive years, Wisconsin. That’s only 112 electoral votes, NBD. Not only are black voters the most loyal segment of the Democratic constituency but they’re less liberal than the Bernie Sanders wing of the party. Should the nominee represent the most active, ideological voters, or should it appeal to the rank and file?

I have for some time thought the Bernie-Trump comparisons were stupid and a misunderstanding of populism and nationalism, as well as how they appeal to their respective bases. I think Bernie and Cruz are in fact partisan counterparts. Both are willing to buck party orthodoxy in favor of ideological purity. Both rile up their respective establishments over concerns of ideological extremity and electability. It’s also true that Bernie is not as liberal as Cruz is conservative, polarization in America is asymmetric. But they make their activists happy, while concerning their party’s rank and file.

To properly set up the nomination process, we need a consensus on what the primary process is intended to do. Neither party has a clear understanding of that process at the moment.



Making Sense of the Democratic Demographic Divide

At this point there’s no question that Hillary Clinton has done an incredible job winning over black voters in the Democratic primary, particularly in the South. Exit polls show her winning black voters almost 4-1 in places like Alabama and South Carolina. The combination of these exits and her huge victories in the former Confederate states underscore this fact. If you’re following this election on Twitter, or even worse Reddit, then the commentary on the Democratic side suggests that there’s a huge divide between white and black voters over Bernie and Hillary. But as pointed out by Corey Robin and others the demographic divide in the South has not shown up in the Northern and Midwestern states with much smaller black populations. In other words there might be a regional effect to the demographic polarization if you will. Look at the exit polls in Massachusetts.


It seems that as states become more homogeneously white (excluding Vermont and New Hampshire for the moment), race becomes a weaker predictor of voting behavior. This of course bodes horribly for Senator Sanders, who needs to win white voters by a big margin the rest of the way to even close the gap with Secretary Clinton, let alone win more delegates.

But in looking at these exits, the really interesting dividing line to me is not race but rather income. It’s not just that higher income people like Secretary Clinton which they do, or that lower income people flock to Sanders. In this primary, and in most, higher income people make up an incredible percentage of primary voters. From the question wording, it isn’t clear whether people making over $100k are reporting individual salaries or household income, but they make up 37% of primary voters, the largest group. For perspective, median household income in the US is about $54,000 and households making over $100k are in the top 20% of income earners. Just in case you thought income inequality wasn’t real. If the exits accurately represent Massachusetts turnout, then almost 70% of primary voters made more than the median household income in the United States. We do have to take into account that Massachusetts is a higher income state, but that isn’t explaining the entirety of the income divide.

It is well documented (Verba, Schlozman, and Brady, 1995) that political participation is a function of resources (knowledge, time), engagement, (interest) and recruitment (networks that mobilize voters like civic organizations, churches, etc) in the US. The misrepresentation in turnout (which is not nearly as pronounced in the presidential election) is in part why candidates like Senator Sanders haven’t been successful in the past. It’s also why the consultant class of Democratic operatives were so hostile to Sanders’s commitment to a political revolution through massive mobilization. They aren’t cynical people who hate democracy or hold contempt for the poor. Nor are they bought buy Secretary Clinton and the big interests. They’re just skeptical of the idea that a candidate or political operation can vastly improve turnout.  There are entire industries within the Democratic apparatus committed to turning out low-income voters to help candidates and referenda up and down the ballot across the country. But making big changes in turnout is an arduous task. In 2008 Barack Obama, a transformational figure who crushed Republicans in two presidential elections helped spur a 2.7% increase in voting age population turnout from 2004to 2008, that was helped by a massive financial crisis and the widespread disapproval of President Bush. President Obama was successful in both increasing turnout and persuading voters his way, but about 40% of the voting age population did not vote in 2008.

These obstacles are not an argument against efforts to increase turnout or better representation of low-income people. The Democratic party will fall apart without it’s working class base. A commitment to those voters is key to their continued success. But if groups that support Senator Sanders do not turnout in places like Massachusetts, which should be favorable turf for the resource-rich candidate, then the populist economic agenda he supports will likely falter. This to me is the most important demographic divide in the party that explains Clinton’s success. She’s doing well with black voters but that shouldn’t be discounted or ignored. She’s doing well with voters. Period. And voters often sing with an upper class bias.

Souper Tuesday Preview: Limited Data

It’s really quite hard to say what is going to happen on a Super Tuesday in which there are many candidates and the race is very competitive. Often the front runner is secured by now and the nomination is not in question a lot like Hillary Clinton on the Democratic side. But given the GOP leadership’s disdain for Donald Trump the Republican race is still somewhat in question, though I think Trump will be the nominee.  You can see from the figure below that there isn’t a whole lot of polling data, which means there could be shifts in the electorate or responses to the campaigns on the ground that changes outcomes in some states.


There was 1 poll in Tennessee this year and it does give us some information about the race, but it’s not exactly foolproof. I love to aggregate polls, so I’d like like to see what the Super Tuesday race looks like at the macro-level.

The reason why I want to depict these graphically is that although I pride myself on my ability to remember top-line poll numbers, it’s 1) not a great use of time or energy, but 2) it becomes an impossible task when you’re trying to differentiate polling data across 10 states with varying number of delegates, distributed using different delegate rules. I omitted Alaska from my Super Tuesday polls because no one has campaigned there and the polling data is almost non-existant. So what are some ways to look at these polls that’s meaningful and helps us understand the state of the race. My first idea was to aggregate. Here are all of the Super Tuesday polls since January 1st.


If you can read that congratulations. It looks like Trump is winning, but it’s a muddled mess. It took an enormous amount of time to make that lol.

Here’s my best attempt at putting that same data in the chart above into one place. This graph is intended to help you see how well the candidates do across the states and allows for comparisons. Some of these states only have 3 or 4 public polls since January 1st.


Click here for the  SuperTuesday chart Expanded:

However,  Vermont and Minnesota only have 1 publicly available poll since January first and Arkansas and Tennessee only have two. I created bar graphs of these states below. Oddly, Kasich was not polled in Minnesota, with him I’d expect Rubio to do a bit worse, which might explain why he’s beating Trump there.



Here’s the expanded version of the barplot above TwopollState

What’s clear from the graphs above is that in the places where we have polls from before the Iowa Caucuses on February 1st, Trump’s support is virtually the same. Winning hasn’t added much to his coalition, though it’s generally grown, but his support has been there for quite a long time, even in places where he hasn’t done that much campaigning. I’ve droned on about how Trump does slightly worse in Live-Interviewer polls than he does in online polls like Survey Monkey, but I’m starting to think that Trump’s support might have even be understated in even the online polls. There’s a sizable portion of  Republican Party primary voters (and it might be nearly half) that want Trump to represent the party. The fact that today is the 4th consecutive day in which the established GOP operatives and candidates have completely gone after Trump to no avail speaks to just how large his support is. One complaint from establishment types is that Trump is only succeeding because of how much media attention he is generating and hogging from the other candidates. However, since his horrible comments about John McCain in July of last year, the majority of Trump’s coverage is negative. Which is to say, his support might not simply be a function of his media attention, it might in fact be sincere support, in spite of the fact nothing but horrible things are being said about him. I think Trump is going to win every state but Texas tomorrow, and I don’t think any candidates are going to drop out. Trump is going to be the GOP’s nominee.

Let me know if you like how these figures look. I spent the entire day working on them, but it’s very hard to graphically represent minimal polling data and have any confidence that they depict the state of the race all that well.

Clinton’s Lead is Narrowing and Growing

One of the weirder trends of the last week that has been completely drowned out by the Trumpening is the asymmetric relationship between Clinton’s state polls and national polls. Although Clinton still has a 7 point lead in the national polling aggregator (shown below) with the exception of Vermont her lead is dwindling and has been since December. However, Hillary is leading in every Super Tuesday state (with the exception of Vermont) by quite a bit. How do you explain what’s going on here?


If I showed you the figure above, you’d assume Bernie was very competitive in the upcoming states.

But look at these state polls from Polling Report’s Twitter feed. Shoutout Polling Report

So these polls in an of themselves aren’t terribly shocking, but if Bernie Sanders is going to win the nomination, he has to win by large margins in Massachusetts and Wisconsin, and be much more competitive in places like Pennsylvania. It’s possible that the national polls are unrepresentative because they disproportionately sample people who aren’t going to turnout. Or it’s possible that the state polls are wrong, though they’re been fairly accurate in the 7 GOP and Democratic contests we’ve had so far. It’s conceivable that Hillary’s support comes from the die-hard partisans who are most likely to turnout in primaries, but she’s less popular with a broader electorate. I don’t have a diagnosis for this asymmetry, but it’s something to watch out for over the next 3 weeks. The national polls might reflect a closer race than actually exists.