Momentum

Trump Winning the Racist Endorsements
Trump continues to earn the backing of the most racist and xenophobic political figures out there. After scoring big wins from Liberty University's Jerry Falwell Jr. and Maine's controversial governor Paul Le Page, Trump has now scored back-to-back "wins" of KKK leader and former presidential candidate David Duke and the French equivalent and former leader of the National Front Jean-Marie Le Pen. It is considered unusual for foreign politicians to get involved in American elections, so the last one is a bit of a coup.

Super Tuesday Updates
Several new polls have come out for Super Tuesday races, and most of them are basically in line with the estimates I gave in the last post. However, this is an election where seeing polls swinging 10% in both directions is considered normal variation. There are some swings in both directions, but there's nothing that surprises me. Of course, this is an election where refined polls and the eventual results only bear a superficial resemblance to each other, so I'm just reporting, not predicting.

Clinton Beats the Spread in South Carolina
Hillary Clinton won decisively in the South Carolina primary yesterday. This win, by itself, should not be news even worth mentioning. Every poll and gut check said that a Clinton loss there would be virtually impossible. What is big news, is that Clinton won by 73.5% to 26%, creating a huge margin that is more than double the spread that was expected. Exit polls show that she won almost every group across age, race, gender and income, leaving Sanders only his stereotypical base of 18-24 year olds and white males. Minorities turned out in record numbers to support her. In Sander's own words "We got decimated." This indicates that the entire South is likely out of play for Sanders, with delegate margins even worse than his campaign was planning for. While it is too early to call this one event "The End of Bernie" (which, yes, many political commentators are wont to do), he will have to beat the current margins on Super Tuesday, or drastically change his strategy. We have two days before we find out.

Super Tuesday

Republican Civil War
I'm hearing a lot of hype about how Trump will "sweep Super Tuesday" or how he could become "the unstoppable nominee" after March 1. Cries of how Trump "has no ceiling" after Nevada, as if fulfilling what we have been seeing since Iowa is somehow shocking. True, no one has managed to take him down yet. However, no one has seriously even tried (parting shots of doomed candidates not withstanding).

The reality is that Trump's support continues to hover at a steady 33-35% nationally, where it has been since the middle of December. The race for the other 64% has been VERY dynamic, however, and may be starting to coalesce into two other candidates in time for the real fight over the primaries. As I wrote before, in order to avoid a brokered convention, Trump would need to at least cross that 40% national support line, which remains out of reach. When we break down the Super Tuesday states, using state-wide polls and the proportional allocation rules of each state, we see this prophesy is being fulfilled.

Due to proportional allocation rules, Trump gets about 43% of the delegates, well shy of the 50% he needs to avoid a potentially nasty fight on July 18th in Cleveland.

There is a lot of assumptions put into this delegate assumption (including how congressional districts split and whether pollsters have any business predicting primary elections), but it gives a good estimate of what we might expect March 1st. However, there are a number of edge cases here that are within the sampling errors of these polls (and well within the de facto margins of error). The delegates are awarded proportionally among the candidates who meet the minimum threshold. If you don't meet 20% in Georgia, you get nothing. Some things to keep an eye on:

• If Trump comes in 4% lower in Alabama, that could be a major blow as he would no longer qualify for "winner-takes-all" at 50%. If Rubio gets gets half of that, Trump would lose 22 delegates to be split among Rubio and Cruz.
• Cruz is close to the threshold in Georgia, potentially costing him 20 delegates.
• Rubio is close to the threshold in Alaska, Tennessee, Vermont, and the very important Texas. This could cost him as much as 75 delegates. That's half of his expected winnings on Tuesday. He needs to make a serious play for Bush's supporters, and some of Kasich's. If you live in one of these states, expect to see ads for Rubio popping up everywhere.

Democrats Get To Vote, Too!

For all of you Democrats out there, you may not realize it from the media coverage, but your party votes on March 1st as well. Like every comparison between the Democratic and Republican primaries, it just isn't nearly as exciting as a three-way battle for the heart and soul and future of the party leading to a possible floor fight at the convention. Sadly, adults conducting business in a respectful manner simply isn't as interesting as grown-up kids throwing temper tantrums. C-SPAN could never compete with Toddlers in Tiaras.

For what it's worth, here's the breakdown on the Donkey Race:

Clinton is expected to win, pretty much across the board on Super Tuesday, racking up 2/3 of the available delegates for a net gain of nearly 200. The only cliffhangers are:

• Will Clinton manage to get past the 15% threshold in Vermont?
• Which states will the pollsters get horribly wrong?

Drunken Polls

The Subtle Joys of Polling Crosstabs
Most pundits are happy to report the top line from a poll, blithely reading off that a candidate is ahead by 20% one day and behind by 15% the next. If there is any internal concern or irony in these reporters, it's long gone by the time the final version is posted. Only on particularly slow news days will you ever hear some tidbits about "second choice options" or other simple cross tabs. However, at Black Swan, it's those buried analytics that are most fascinating to us, even if we realize that sane people consider the inner workings about as interesting as webpage source code.

Fortunately for you, we have found an intriguing set of cross tabs that I think you're going to love! (If you are still reading at this point, thank you for your faith in me, or just your morbid curiosity.) Public Policy Polling (PPP) has surveys that go deep into the mind of likely voter to find their first and second choice for the Republican nomination, as well as several match-ups in two-way or three-way races. The results are fascinating, and not from a strictly political sense.

Is that Jim Gilmore Supporter Drunk, or Just Stupid?
No, this is not a slight against Jim Gilmore, or anyone who thinks that he would make a good president. This is about one specific person in New Hampshire who was contacted by PPP. I can talk about just one individual, because there is only one person out of 515 Republicans who supports Gilmore (okay, maybe that was a slight against him).

This one supporter, we will call them Gil in honor of the candidate, is easy to track through the poll, and see their various answers. For instance, Gil identifies as a moderate Republican, who is male and over 65. Of note is that when asked to name a second choice, Gil prefers Ted Cruz. However, in a 4-way race between Cruz, Rubio, Trump, and Bush, Gil changes his mind to Rubio. Then again, if you remove Bush from the contest, to a Cruz-Rubio-Trump contest, he's back to Cruz again. Switch the 3-way to Cruz-Rubio-Bush, and Gil likes Rubio again. Hey Gil, you do know that you can still vote for Cruz, right? Alright, clearly this guy likes Rubio and Cruz, but which one does he like better in a head-to-head match up? Ted Cruz. So if the race were just Cruz vs. Trump, you would vote for.... Trump. Of course you would, Gil. Hey, can you tell me how many fingers I'm holding up?

This goes on and on, with this one loon contradicting himself every chance he gets. It turns out that when asked if he has a favorable or unfavorable opinion of each candidate, he doesn't even like Jim Gilmore. So clearly this guy is unreliable, and probably just messing with the pollster. You're bound to have a few outliers who are adding just a little extra to your margin of error.

Let's switch to Iowa, where things are bound to be better, right? Things are looking good for the Gilmore campaign there, as PPP found TWO Gilmore supporters out of the 510 likely Republican caucus-goers surveyed. On another positive note, only one of them has an unfavorable view of Big Jim (the other supporter still isn't so sure). Also, both of these supporters belong to a critical group of voters that the Republicans would like to win over in the general election: liberal female tea-party members. Jim Gilmore does seem to attract the most interesting people.

Cruz Supporters Only Slighter More Sober
It's a good thing that there aren't many of those crazy people responding to the pollsters. For instance, Ted Cruz is a serious candidate with serious popularity, and you can count on his supporters not to be messing around like Jim Gilmore's. At the very least, 100% of Cruz's supporters in both states view him favorably. However, it does kind of beg the question of why it is that, just two minutes later, when the pollster asks New Hampshire about the four-way match up between Bush-Cruz-Rubio-Trump, only 83% are loyal enough to Cruz to pick him a second time, while 13% defect to Trump. That's roughly seven different people out of the 55 or so Cruz supporters contacted. It's ok, though, because in a two-way race between Trump and Cruz, most of them jump back into Cruz's camp. They do seem to lack a certain courage of conviction.

Rubio doesn't have it any better. For any given match-up involving Rubio as an option to vote for, anywhere from 12-15% of his supporters in New Hampshire choose another option. (The one exception is a head-to-head against Trump. Apparently there are some things you just don't joke around about.) In Iowa, that number for Rubio is closer to 17-20%. In another camp, Bush can hold on to about 90% of his supporters through any match-up in NH, and between 82-88% in IA. Trump is apparently the only candidate who has a strong cadre of no-nonsense supporters, maintaining 95-98% no matter how many alternatives they are given.

America is Sick
Across the board, at least 10-15% of the voters surveyed give weird and contradictory answers over the course of a single survey. Keep in mind that these are the likely voters, according to the experts' models. So it's not like they would lie, or not know who they were voting for, right? Instead, I have a few hypotheses about what is happening:
a) These respondents are severely inebriated in some way
b) Contrary to earlier reports, Iowa is not a native English speaking state
c) There is rash of under-reported concussions sweeping New Hampshire
d) Bored respondents are just handing the phone to their two-year-old child/grandchild to play with
e) The most jaded political pundits are right, and the American electorate literally does have the memory of a goldfish

No matter what the cause is, it should be clear that there is an additional source of error that goes far beyond the mathematical sampling error that is being routinely reported and used to justify these results. Furthermore, this previously unreported "bat-guano-crazy error" is somehow making it through the "likely voter" screens (which apparently don't include questions like "Have you been drinking this evening?"). So the next time that you see a headline about a candidate jumping 20 points in the polls, consider what the people they polled might have been smoking.

(For those interested in seeing more of the insanity, you can look at the New Hampshire and Iowa polls directly.)

The Dark Horse Surge in New Hampshire

So, uh... what's up with New Hampshire Democrats?
The news of the hour, on the Democratic side, is apparently Sander's stunning surge to a 27% lead over Clinton in the latest CNN/WMUR poll in New Hampshire. Pretty much every political outlet and pundit that I have seen is quoting this number, as if the Oracle handed down the result at Delphi. I would like to include it as an interesting counterpoint to the other polls taken in that state in the past week.

It is pretty quickly apparently that one of these polls is not like the others. One of these polls just does not belong. The ARG poll (which is technically more recent) shows Sanders leading by 6 points. Admittedly, +6 is still a lead, but that's a huge difference from +27, and is in line with their previous two polls in the past month. The Monmouth poll of +14 gives the most similar result to CNN/WMUR, but is still off by double. Public Policy Polling (PPP), which performed decently well in 2012 and 2008, even gives Clinton a consistent +3 lead.

Now I'm clearly not the type of person to conclude that the minority opinion is wrong for being in the minority. Normally, I would look at three factors to validate a poll result: the demographics, the pollster's past accuracy, and the trending numbers from pollster's past results. Unfortunately, the poll did not include anything even vaguely resembling a demographic in their report. (If a more detailed report does exist, please send it to me!) The pollster (WMUR) is pretty much held to New Hampshire, but in 2012 they were highly erratic with their predictions, with back-to-back polls just days apart showed swings of 9 points. Finally, this recent result represents a large shift from December (50% to 40%) and September (46% to 30%) neither of those results or their relative motion were supported by any other outlet. Conclusion? WMUR, you're drunk. Go home.

So, uh... what's up with New Hampshire Republicans?

On the other side of the coin, something unexpected seems to be happening. There are now unconfirmed reports that John Kasich may be running for president. You may be excused if you are currently scratching your head trying to figure out who I'm talking about, but if you rewatch the Republican debates, you will distinctly see his name show up in the credits.

While desperate reporters and pundits are trying to put together a story about an outlier poll showing an upstart surging ahead in New Hampshire, they completely missed my favorite one: ARG reports that John Kasich is the anti-Trump, running only 7 points behind the "buffoon."  Plus, this exciting story actually has some potential meat behind it, with recent polls from Monmouth, Reach, and ARG (again) supporting the missing link of Kasich's surge to a weak second place in early January. Now none of these polls are exactly gospel, but it is now highly plausible that Kasich could easily take second place and the self-proclaimed "Prince of Light and Hope" could emerge as the new establishment pick. With 49% of NH primary voters making up their minds in the last three days before the election, combined with an electorate that isn't completely bored with hearing his name every day, he has a lot of room to change some minds and mount a successful surprise attack.

Prediction: Cruz Wins Iowa Caucus

Prediction: Cruz Wins Iowa Caucus
Based on the latest poll data, it is likely that Cruz will win the most delegates in Iowa. The current estimate of the popular vote among caucus-goers is Cruz: 30%, Trump 22%, Rubio 14%, with remaining candidates' support being too small to be certain. The 95% confidence is +/- 8%, representing the high uncertainty of this type of race.

There are three key factors that I am considering:
1) The result of a couple of polls with high quality samples, which come close to the exit poll demographics from 2008 and 2012 (Fox News and NBC/WSJ/Marist)
2) The data from a PPP poll, giving some insight into the second preferences of voters
3) An examination of past behavior of Iowa caucuses

One important factor to keep in mind is that approximately half of the people who attended the caucuses in 2008 and 2012 did not make up their mind until the last month, with about a quarter deciding in the last week or less. This is supported by the polls that I can see from those times. This doesn't mean that they had no idea who to support, or that they wouldn't end up going with their best guess from December. However, it does mean that the majority of voters' preferences are still fluid, and they could easily change their mind to a second choice.

This is important as we look at a number of candidates at the back of the pack, who may be seeing falling numbers. As certain candidates seem less viable, a snowball effect takes place, where falling support makes voters who are leaning towards that candidate to re-evaluate their position. The very nature of the caucus amplifies this effect, as an uncertain voter walks into a person's home or other small gathering on caucus day, and finds that they are pretty much alone in their support for candidate A.

So where are these fluid votes going to flow? For any candidate looking at less than 5% of the vote, the second choice is for Cruz over Trump, at more than 2:1. Particularly noteworthy is Paul supporters, who are know to be well organized and well placed for the caucuses, who favor Cruz 5:1 over Trump (but with almost equal preference for Carson).

Another interesting point is that while voters who lean towards Trump are still considering Cruz as a second choice (36%, far above any other candidate), Cruz supporters are much more likely to support Rubio than Trump (26% to 18%). Similarly, 36% of Carson supporters see Cruz as a second choice, versus 12% considering Trump. The bad news for Rubio is that, other that Cruz supporters, almost no one else sees him as a good second choice (including Bush supporters).

The take home message on this is that in the midst of a chaotic field, any volatility aims to favor Cruz.

The Importance of Election Day Turnout

The 'Voter Turnout' Strategy
There's a lot of talk about turnout being the deciding factor in the coming election. According to the theory, the number of "true independents" or "swing votes" is small and getting smaller. Partisan politics is driven more and more by identity, and the dominance of party-line voters is the norm. The implication is that it isn't worth the effort to try to win votes from the "other side" since Democrats won't support a Republican any more than a Red Sox fan will cheer for the Yankees. Instead, the strategy says that elections are won by getting more of your team out to the polls than the other guy.

The politicians' belief in this theory does help explain the hyper-partisanship seen in today's campaigning and legislation. It also seems to drive the extremism we are seeing from the different candidates, putting the two parties further and further away, ideologically. For instance, Sanders and Clinton debate how much to increase the minimum wage, while the Republicans are trying to decide whether the existence of a minimum wage is unconstitutional or merely a very bad idea.

Validate the Theory
Given how much time is spent by the pundits talking about these points, very little ink and bytes has been spent looking deeper into this effect. For an empirical evaluation, let's turn to the past presidential exit polls to see how much truth there is to this point.

In 2012, 92% of Democrats voted for Obama, while 93% of Republicans voted for Romney, while self-declared Independent/"Other Party" voters split 50%:45% for Romney. Given that 32% of Americans identify as Democrat and 30% identify as Republican, you might be forgiven for thinking that Romney should win the popular vote, 49.5% to 48.5% (with 2% to "other" candidates). However, the actual turnout for Democrats was 38% of voters, vs. 31% for Republicans, with a final tally of 51% for Obama to 47% for Romney. A disproportionate turnout for Democrats clearly was the deciding factor in an election that Obama "should" have lost.

While this phenomenon is not an aberration, it is a relatively recent pattern. In 1980, Reagan came to power amidst a strong turnout for Democrats by winning over voters of all backgrounds. After that, we see Bush the 1st and Clinton (the 1st?) win elections by reaching out to the other party while voter turnout remains stable. In 2000, Bush the 2nd inherits the same voter turnout distribution, and wins the electoral vote despite narrowly losing the popular vote.

It is in 2004 that we first see the triumph of "turning out your base", with a surge in turnout for Republican and conservative independents providing the deciding factor in what otherwise could have been a nail-biting repeat of 2000. Whether intentional or not, the Democrats learned their lesson, and countered with increased turnout of their own, plus a large support from the independents, to win in 2008. However, in both 2004 and 2008, the winner of the popular vote (and presumably the electoral vote) would have been the same without it.

The conclusion: maybe voter turnout isn't the definitive strategy it's cracked up to be. It's a great way to tilt the tables a little, but a strategy of winning over the hearts and minds still has a better track record. Now, we continue to hear how the middle is shrinking and swing voters have gone the way of the dinosaur, so perhaps more than four years is too far back to look in politics. If the Republicans actually nominate Trump or Cruz, we will see if that is really the case.

Divergent Polls

Divergent Polls

Last week saw a rash of polls released. Here's a quick summary of those numbers:

This is a good crop of fairly consistent polls, all within the expected sampling error of each other. From these polls, we can conclude that Trump is on another major surge, reaching for that magic 40% range he needs to avoid a brokered convention. Meanwhile, Cruz is showing minor gains and Rubio appears to have stalled.

The most interesting thing, however, is not in these numbers. It's in these other numbers:

This is the same race, across the same nation, during roughly the same period of time. These polls tell a very different story, where Trump is struggling to expand his support beyond the 28% ceiling we've seen for months. Cruz has launched a major surge towards the front-runner slot, and Rubio's growth continues at a slow but steady pace.

Two Sets of Polls, Two Sets of Reality 

This is not a case of an outlier poll, like most political stories seem to be casting it. The odds of the generating the Quinnipiac poll as a statistical outlier from the first grouping is roughly 1 in a quadrillion less likely than the Fox News poll. That's 1 with 12 zeros behind it. If that isn't unlikely enough, think about the odds of that happening three times. 

This means that these are two very different modes, which are clustered within sampling error and are self consistent across multiple candidates. That's two different answers for what is happening, and they cannot be treated as polling the same reality.

What the *!@? is Happening Here?

The big question is what's so different about these two groups? For once, it's not cell phones. Both groups are using a similar mixture of landlines and cell phones with live interviews. Also, it's not registered voters vs likely voters, or live calls vs robocalls, or any of the other usual suspects. Looking at the other methodologies doesn't show any other tell-tale signs. The truth appears to be more subtle than that. 

Unfortunately, most of the pollsters don't like to reveal any of the deep internals of their models, or report any of the raw results. However, a deep dive into the cross tabs does show something interesting. I performed some statistical forensics to back out magnitudes of how many minorities were interviewed in the poll, inferred from the statistical variation between different categories of registered Republicans, such as "All Males" vs "White Males." Keep in mind that Gallup polls and exit polls from 2012 and 2014 put the number of minorities who are registered or self-identifying as Republicans somewhere around 11-13%. If these groups are interviewed at this proportion from the general populace, even on non-racial or non-political issues, we would expect to see a certain amount of differences solely from sampling variance. As the size of the group goes down, so does the variance. In short: if you think white people and minorities agree on everything down to 1%, you didn't bother to ask one of the groups.

For all of the first group of polls, the number of minority respondents appears to be very low, bordering on non-existent. Across all categories, the variation between these two groups is never more than one percent. That includes estimating Trump's support. If minorities seem to support a racist candidate like Trump as much as white people, there's something strange going on. 

In the second grouping, computing the crosstabs is a little different. Two of the pollsters (Suffolk & Quinnipiac) are university polls that flat out tell us how many minorities they reached. While they didn't break it down by ethnicity within the Republicans, both pollsters hit very close to the target number of Blacks without weighting, and did a good job of representing Hispanics and Asians with only a little undersampling.  Furthermore, they seem to give a good natural distribution from the very low income groups (<$30k/year) and younger age groups than the first grouping.

I dug a little deeper, and looked at results for the first group, going back three months. This is roughly the time period where we have seen the bifurcation of the polls that has so recently accelerated. It is also during this time that the undersampling of minorities seems to take place. The further back in time you go, the more "Whites" seem to differ from "All Republicans."

Conclusion A: Quality of Sample Matters

While excluding minorities isn't enough to explain the difference in polls by itself, it is a clear indication that the quality of the sample is bad for roughly half of the major pollsters. No amount of re-weighting will account for having only three Hispanics answer your poll. Also, this consistent degradation of the quality likely applies to other subgroups as well, since it is highly doubtful that these pollsters are just being racist somehow.

Conclusion B: Trump Might Be In For a Rude Awakening

As it stands, I could believe that many polls are not only suffering from higher than reported margins of error, they are also inadvertently introducing a consistant bias into the results. I caution readers to start treating the first group of pollsters with a certain amount of skepticism until February 9th (New Hampshire's primary) when we have a little more information to go on. The second group of polls, which show Trump stalled and Cruz rising fast, seem to be much more believable once you look under the hood. 

It is quite possible that Trump's 'lead' is still at -22%.