On Nate Silver and 538’s 12.2% Clinton House Effect

cross-posted at CounterPunch

Nate Silver has taught us all a lot of things about how to accurately assess good and bad polling and how to forecast difficult to forecast political races in the United States accordingly. Unfortunately, he and his web juggernaut FiveThirtyEight dot com have forgotten several of their own lessons during this 2016 primary cycle.

And it shows.

It isn’t just that Silver and his main sidekick Harry Enten badly missed that Donald Trump’s early poll numbers were real (Enten famously said Trump had a better chance of playing basketball in the NBA than he did of securing the GOP nomination). But, outside the South, 538 has a 12.2% Clinton House effect on the Democratic side. In eighteen of the twenty-one races, 538’s projection or modelling has shown a Clinton bias.

It gets even worse if we ask about the three of the twenty-one. One of the three was Vermont. Not exactly something to write home about, having a relatively slight pro-Bernie bias in Vermont. If we suggest that calling a race within three points means you actually got it right, with an inevitable slight bias one way or another, then Ohio, at a 2.8% Sanders bias says more, perhaps, about an under performance there. Finally, there is voter suppressed Arizona that is starting to look like Florida 2000, and where it isn’t particularly easy to peg what FiveThirtyEight thought about it ahead of time anyway. Their main stance toward the Democratic race, most charitably interpreted, has been one of paternalistic boredom. It showed in Arizona. More on that in a bit. Since the only argument left on the Democratic side is Math! (think Jeb!): To The Numbers!

538 Bias Chart Final

You’ll notice methodologically that I had to account for places where there aren’t Poll Plus FiveThirtyEight projections. This was quite simple. In a piece entitled, Bernie Sanders Doesn’t Need Momentum – He Needs to Win These States, Nate Silver projected what each remaining state would look like if the race were tied nationally (If, if, IF “but it’s not” has been the constant refrain). So, if the race isn’t tied nationally, there is no way Sanders should be outperforming his projected numbers for those contests, right?

Read More


Projections for Sanders v. Clinton March 22-26

All in all, I did mediocre or worse at projections for the Ides of March contests between Hillary Clinton and Bernie Sanders.

I nailed the outcome in North Carolina exactly (13.8% victory for Clinton, where RCP had Clinton +24.0 and FiveThirtyEight’s Polls Plus projected a 24.6% Clinton victory). I also correctly called Sanders winning more than 44% of the vote in Cook County (Chicago), Illinois. He took 45.6% of the vote there. Cook County is majority non-white alone (24% Black, 24% Latina or Latino, 43% white alone). This in and of itself should interrupt the “Sanders can’t compete in diverse places” narrative. It hasn’t.

Nevertheless, I was wrong by 7.1 points in Florida, 7.6 points in Illinois, 9.6 points in Missouri (also wrongly calling who would win), and 12.0% in Ohio. In each case, I was bias toward Sanders. My average for last Tuesday was a 5.26% Bernie bias.  I signaled my model couldn’t really accurately predict Florida. I also said that Sanders likely had a Kasich problem and had short-shrifted Ohio in competing for Florida. Both were true. He also had a Spring Break at The Ohio State and Cleveland State problem. And Ohio is one of a tiny number of states that has never voted against a Clinton where it had a chance to vote for one of them. I stated that I wouldn’t be surprised by a 10 to 12 point Sanders loss in Ohio. The Facebook Primary model actually suggested that, as it did a very slight win for Clinton in Illinois. It would have predicted a very slight win for Sanders in Missouri.

I would have done much better if I had simply stuck with the Facebook Primary model and called 35-40 points for Clinton in Florida, 10-15 for Clinton in Ohio, 15-20 for Clinton in North Carolina, and races too close to call in Illinois and Missouri. So what does the Facebook Primary, adjusted to USPD, tell us about the next six races between today and Saturday?

Utah, Idaho, and Alaska look even better for Sanders than similarly situated states Oklahoma, Nebraska, and even Kansas. My projections for them run accordingly. Hawai’i and Washington look even better for Sanders than similarly strong Blue leaning purple state Minnesota, with Washington in a stronger position for Sanders than Hawai’i. Arizona looks very similar to Nevada and Iowa with a slight nudge toward Sanders (details in note below). Arizona is the toughest to call and the most important contest of the six. In the other five, he should win by an average of much more than the now well-known 58% of delegates he needs in the remaining races. 58% is likely his absolute max win in Arizona, but he could also lose the state, perhaps by as many as five to ten points or more. Interestingly, it seems as if Sanders’ share of the Latino vote is tracking very closely with Obama’s in 2008 (Nevada being the exception). Clinton won with Latinx voters 56%-41% in Arizona in 2008. With likely as much as 30% of the Arizona Dem primary being Latina or Latino according to a press release from SVREP, if these are the numbers again, Sanders will likely lose decisively. My projections here are hewing very very closely to the Facebook Primary, adjusted to USPD.

State    Spread                  Delegates
Utah +35.8 Sanders        HC 9 BS 24
Idaho +25.6 Sanders       HC 9 BS 14
Arizona +0.8 Sanders      HC 37 BS 38
Hawai’i +24.9 Sanders     HC 9 BS 16
Washington +40 Sanders HC 30 BS 71
Alaska +39.4 Sanders      HC 5 BS 11
Clinton 99      Sanders 174

Read More

Now Is the Time to Win Against Toronto Police Violence

Still Frame from Video by Vice Reporter Jake Kivanc https://youtu.be/0OHcNO2wtN4

Still Frame from Video by Vice Reporter Jake Kivanc https://youtu.be/0OHcNO2wtN4

Now is the time to win against Toronto Police violence. The right pieces are in place. The right demands are being made. The right people are at the centre of the right battle. And, they are using tactics and demands that will work.

As many of you know, I’ve spent the better part of the last decade pushing back as hard as I know how against Toronto Police violence. Usually with very little success. I am very pleased to see that people I have met along the way like Ellie Adekur (right in the thick of it on the ground) and Desmond Cole (with his beautiful pen) are helping to lead the charge. It helps enormously that they are people with in-depth analyses of their own potentials for misuse of power but who nevertheless ACT.

I hope that any friend of mine who is still in Toronto and able will join the resistance in any way they can while being incredibly mindful of the need not to get in the way of African Canadians having complete charge of their own path to liberation. Just assume that they will take some decisions and engage in some actions that you disagree with. Have the grace to recognize that you may well be the one whose ideas are off or, even if not, that giving people space to screw up on their own is critical to recognizing and honouring their full humanity.

The Ides of March Is Upon Us: Will Bernie Sanders Survive?

cross-posted at CounterPunch

Bernie Sanders needs around a 320 delegate haul in today’s Ides of March contests in Ohio, Missouri, Florida, North Carolina, and Illinois to stay viable as a candidate at least through New York in late April, if not all the way through to California.

In the six contests since Super Tuesday, Sanders will have garnered about 141 or 142 delegates (five are still to be assigned) where my Narrow Path to Victory suggested he would need about 144 in that time frame to stay viable. That same map suggests that Sanders should get 330 delegates today to continue along the Narrow Path. If Sanders falls below 300 delegates today, I will begin agreeing publicly with the insistence that the math is nearly impossible for Sanders to overcome. Above 320 and there is little doubt in my mind that Sanders could very well find his way to a final show down in the Golden State. The no-man’s-land between 300 and 320 will be more difficult.

My record of seeing things accurately in advance since the blowout in South Carolina is fairly good, particularly over the last week and a half where I nailed Michigan and only really missed just how big Sanders would win in Kansas (see note below for more detail).

So how do I see today going down?

I’ve had a bit less time for in-depth detail, but have given some initial thoughts on Ohio, Florida, and Illinois. Sanders has very much created Michigan Magic in Illinois, and I am predicting that he will win there even bigger than in Michigan. Ohio and especially Florida are very difficult to call. For Florida, I will be giving ranges and what they would mean. The basics (??? next to Florida because I have used close to the lower end of my larger predictive range as a placeholder in order to see a reasonably possible delegate count):

BS = Bernie Sanders HC = Hillary Clinton

BS = Bernie Sanders
HC = Hillary Clinton

[State by State looks below the jump if you are entering from the homepage.]

Read More

Four Factors That Make Clinton’s Margin in Florida Too Tough to Call

As every other poll, poll aggregator, or armchair prognosticator of any interest has forecast, Hillary Clinton will beat Bernie Sanders in Florida tomorrow, almost certainly by a margin large enough for the networks to call immediately at poll closing.

Four factors, however, make the actual margin of victory too difficult for me to make a reasonable prediction. The final days of polling have seen five pollsters cluster in the 25-28 point range, and this is certainly a reasonable possibility, but the polls on the Democratic side of been wildly horrible at spotting important demographic outliers and late breaking trends in contests from South Carolina (and even Nevada) through to Michigan. Here are the four factors:

1) Age and the Facebook Primary – I have been using FiveThirtyEight’s Facebook Primary numbers adjusted to UPSD to forecast election results with reasonable accuracy even where other forecasting models, including scientific polling, are wildly off base. If accurate, the Facebook Primary would suggest that the 25-28 point range, if anything, is too favorable to Bernie Sanders, with a 30-40 point loss quite conceivable. I do not, however, think comparing Florida’s mostly elderly population to Facebook usage across the United States is likely to yield significantly accurate results the way it has in other circumstances.

2) Age and Millennial Turnout – While pollsters are well aware of Florida’s over fifty population regularly making up three-quarters or more of the electorate, it is not clear that they get the phenomenon that is making the percentage of millennial voters much higher state by state that 2008 in virtually every contest to date. Of course, Obama versus Clinton inspired new and young voters in record numbers, numbers which are not necessarily bested by Sanders versus Clinton in raw format. That said, the overall turnout for Democratic primary voters is down by around 30-35% in nearly every state so far, meaning that 18-29 year old voters, especially, are making up a much greater percentage share of the vote.

3) Latina and Latino Vote – Numbers about how Florida’s substantial Latinx voting population will go are perhaps harder to predict that anything else ranging, in reputable polling, from a 60 point margin in favor of Clinton to a basically even race with a slight edge for Clinton. How many Hispanic voters turn out, especially millennials, and how they vote could make a very substantial difference in how many delegates each candidate winds up with.

4) Jewish Vote – Isn’t it at least possible that the first serious Jewish candidate for a major party could pull unexpectedly high numbers from the approximately 10% of Jewish voters who will turn out tomorrow in Florida? Has anyone seriously entertained this question?

Again, more in my final posting forecasting tomorrow’s results (but I won’t give an exact percentage for Florida, most likely).

Bernie Sanders’ Kasich Problem in Ohio

Bernie Sanders’ narrow path to the nomination has a John Kasich problem. It is simple, measurable from multiple vantage points, and quite worrying if you hope to see Senator Obi-Wan through to the general.

John Kasich is eating Bernie’s lunch with Independents in Ohio.

[This is a first draft and will likely be updated late this afternoon or early this evening.]

Per the larger model I have been using (i.e. including but not limited to Facebook Primary adjusted to UPSD), Sanders could and should do somewhat better in Ohio than in either Illinois or Michigan. NAFTA has long been more openly hated in Ohio than Illinois. Purple states go better for Bernie than Blue states like Illinois. Voting a week later than Michigan gave the Sanders campaign more time to fine tune the anti-trade deals attack. And the overall share of African American voters in Ohio is a bit lower than Illinois or Michigan.

I started noticing a few weeks back, however, that Sanders does not do as well as I would expect in the Facebook Primary in Ohio. It is not that he does particularly worse vis-à-vis Hillary Clinton than expected, just slightly but measurably. Rather his overall share of “likes” in Ohio is much lower than in similarly situated purple states. It appears that Kasich’s understandable dominance in Ohio’s Facebook Primary is attributable to reasonable Republicans and Independents going big for him.

Theoretically, this would hurt Clinton and Sanders equally: Independents who might otherwise vote in the Democratic primary tomorrow will instead vote Republican leaving Sanders and Clinton to compete for a small share of Independents and a larger share of Democratic partisans. But since Independents have gone very heavily for Sanders in prior contests, this could become a major factor Tuesday. Read More

Illinois: Michigan Magic or Final Disaster for Bernie Sanders

Hillary Clinton and Rahm Emanuel (Associated Press)

Hillary Clinton and Rahm Emanuel (Associated Press)

Michigan Magic is still possible in Illinois for Bernie Sanders, but unless he can afford to spend two and a half of the final four and a half days of the campaign barnstorming the state, Hillary Clinton could win by 10% or more.

Senator Sanders has built small fires in the South Side of Chicago with solid black and Latino organizers and even down state with a shrewd visit last week to Southern Illinois University, but unless he can turn those and a modest blaze on the North Side into a raging, ten alarm fire between now and Tuesday, Clinton could manage the convincing victory north of the Mason-Dixon line she needs to declare final victory.

Using the same methods I used to correctly project an epic upset for Sanders in Michigan, including forecasting the outcome rightly in 89% of the counties, I have looked at the basics in nearly every county in Illinois. (I do not, however, have the same insight into Illinois politics that I do into Michigan, where I am registered to vote. I have lived in Chicago less than three years while having been married to a Michigander for fifteen years.)

Cook County (Chicago) and the “collar” of five high-population counties surrounding it is, of course, of incredible importance, but it is not everything in Illinois politics.

Sanders won in Michigan by making serious headway with black voters in Flint and Detroit while winning very big nearly everywhere outside the Saginaw, Flint, Detroit corridor. Smart, well-timed visits in the final week to Western Michigan and even a quick touchdown up north in Grand Traverse County meant the difference between a narrow loss and a narrow victory in Michigan. The same type of strategy can pay important dividends in Illinois, but there are five states in play instead of one this week. And the geographical spread in Illinois is fundamentally different than Michigan’s. The vast majority of Michigan’s African American population is in Detroit, 88% black. Illinois’ overall African American population is similar (very slightly higher) than Michigan’s but is more distributed with Chicago having a 37% African American population.

Adding an additional factor is the much more substantial Latino population in Illinois. Sanders’ enthusiastic support from Chuy Garcia, who had a great showing but ultimately lost to Rahm Emanuel in the last mayoralty election in Chicago, can go a long way, but it would go further if Sanders and Garcia were making multiple in-person appearances at events together.

Face to Face Politics Matters!
A quick look at Kent County in Michigan versus the four mid to large counties my model got wrong in Michigan (Saginaw, Macomb, Oakland, and Berrien) tells us almost everything we need to know. Each of these five counties, Kent included, has African American populations of 10% to 20%. My model, based primarily on weighted use of FiveThirtyEight’s Facebook Primary, predicted victory for Sanders in Saginaw, Oakland, Macomb, and Berrien, but he lost each of them by between one and six percent. While all but one of these four counties is clustered near Detroit and Flint, they did not receive the specific campaign attention with Bernie Sanders headlined events that Detroit and Flint did. In the Western half of the state, Berrien County did not receive a specific visit from Sanders and he lost there by five percent where the model predicts a victory. Kent County, with a 10.4% African American population, did get a specific visit from Bernie and went for him, as the model would predict, by a very large margin (30%).

Before a quick look at Chicago, then, here are the counties where a specific visit from Sanders could make a very big difference (of course, there is no way with the other states in play that he could visit them all, choices must be made):

Sangamon (Springfield) – currently too close to call for the model. Bernie has an advantage in the Facebook Primary, even adjusted by USPD, but it has a 12.4% African American population. Will it vote like Kent County or like Berrien? A visit to Illinois’ capital could mean the difference.

Peoria and Madison – these two counties are clustered close to the St. Louis region and a 1/2 day in St. Louis, 1/4 day in Chicago, 1/4 day in Peoria split, well-executed, could hit all the right marks. Peoria is 18.1% African American, 4.6% Latino, and has a population of 187,000. Madison is just 8.2% African American, but with a population of 269,000 and multiple universities, Sanders is underperforming what he could and should be according to my adjusted Facebook Primary model.

Collar Counties – the five, very high population counties surrounding Chicago are each large and have diverse factors in play where Sanders visiting or not visiting could make a big difference. It is impossible for Sanders to visit all of them. If I had to choose just one, a major event in Kane County (Aurora) featuring Sanders and Garcia would be most important. Kane’s population is 6.1% black and over 30% Latino.

Winnebago (Rockford) – On the Facebook Primary map, Winnebago looks a lot like Genesee (Flint), but it has a 12.9% African American population compared to just over 20% in Genesee. I would say the race is too close to call with a slight edge for Clinton at the moment.

I may update this post with more information in the next few days, but let me just say this quickly:

Sanders versus Clinton in Chicago is actually the undercard.

The main event is the people versus Rahm Emanuel with the election of Cook County State’s Attorney the epicenter of everything political in Chicago right now. Kim Foxx is the challenger and an ally of the most popular politician in Chicago, Toni Preckwinkle. Rahm Emanuel’s partner in crime in the contest is the incumbent Anita Alvarez. Hundreds of Chicago activists who don’t actually care much for either Sanders of Clinton have made this election their reason for existence since the end of street protests after the release of Chicago Police’s Laquan McDonald snuff video. How they, and the people they turn out, vote on Sanders versus Clinton (if either) is a massive wildcard. It looks like the Sanders campaign may have an event scheduled with Kim Foxx on election eve. If this event comes off and how it plays out is anyone’s guess.

If the election were being held today, I would make a call of Hillary Clinton by 10%. Bernie Sanders visit to Chicago State a few weeks ago was the right move, but it is not nealryenough on its own, especially with an very late start to a significant organizing ground game.

As a Sandernista, I am incredibly glad that the campaign has made such bullish moves in Florida. It’s time for some major Illinois love, however. Give North Carolina and Missouri a reasonable chunk of attention, in my opinion, make a huge investment in anti-trade deal commercials everywhere all throughout Ohio and spend a day plus a bit there, touchdown once more in the Sunshine State, but it is all out for Illinois or bust.

CNN Poll of Suburban Senior Citizens Predicts Clinton Landslide in Ohio


This is a stub post. It will be updated later should I choose to give this kind of schlock the full-throated mockery it deserves. If you want real polling for Dems in Ohio, consult Baldwin Wallace or Quinnipiac. (I am quite skeptical of Quinnipiac in Florida via this same press release but haven’t gone in depth with any of the polls or the lower levels of the Facebook Primary. The info available here for Ohio is very thin, but if Sanders is to lose it, 9% is a reasonable floor based on my model.) Also, I was about 89% accurate in picking which candidate each county in Michigan would go for, not to mention getting the topline almost perfectly right. I’ll do the same for at least Illinois and Ohio ahead of March 15.

Screen Shot 2016-03-09 at 9.02.08 AM

Source: http://i2.cdn.turner.com/cnn/2016/images/03/09/reloh1ohio.pdf

Michigan Proves Bernie Can Win; Beware the Ides of March

As predicted by me here, with some noteworthy exceptions down at the county level, Bernie Sanders won a stunning upset in Michigan and, contrary to the “he still can’t win” spin, proved that he can stay in this thing with a reasonable chance at victory all the way through June 7.

All told, it’s been a stellar run since Saturday with Sanders meeting or exceeding my delegate expectations, it appears, in every state except Mississippi. The excess delegates in other states, especially Kansas, should be enough, when everything in Maine and Michigan is finally assigned, to mean he keeps pace with “Bernie’s Narrow Path to Victory.” (Remember that my delegate math is different than FiveThirtyEight’s primarily because it is based on a less-rigidly maintained anti-establishment campaign narrative rather than The White and Liberal™narrative).

But beware the Ides of March.

To put it crassly, Bernie needs five Michigans on March 15 and he only has one week to do it.

Ok, that also isn’t quite accurate. The Sanders Campaign, as noted previously in “Bernie’s Narrow Path,” does not need to win in Florida (214 pledged delegates), but he cannot afford another thirty point loss as in Texas. As it is the state where my two oldest children were born, it warmed the cockles of my heart when I realized late last week that Bernie also is not conceding North Carolina (107 pledged delegates). My model suggests that Sanders needs to garner about 130 of those 331 pledged delegates, or about 40%. There is good news to be had in North Carolina based on polling by Elon University (note, Clemson got South Carolina right and Michigan State got last night nearly right where all the other pollsters or poll aggregators were way off), but it will require all out effort to meet the potential. Florida has a few real wild cards that are simply too much to predict rightly. (More when I address and attempt to forecast reasonable ranges for each state in the next few days).

The other three states, however, require definite wins from Sanders. Once again, there is awful polling for Sanders in two out of the three states (Ohio and Illinois). I need to get into the weeds with the numbers there, but I suspect they’re equally ridiculous in terms of age breakdown, polling size, use of landlines only, and inability to fathom how rural voters will go. In Ohio, at least, we once again have a University poll that RealClearPolitics has perversely refused to include in their accounting that says Sanders was already leading by one point as of February 20th. In Illinois, we are stuck with nonsense like this side-splitting idiocy from Chicago Tribune yesterday suggesting a 42 point deficit for The Bern. Nope.

The third state Sanders needs to win is Missouri, where we have no reliable polling, so far as I know, since August. Sanders needs to take about 200 delegates out of the 370 available between the three (about 54%) to stay viable and really ought to be closer to 225 (61%) so that he isn’t storing up wrath for himself in states like Pennsylvania and California where my model predicted much better outcomes as the Sanders campaign finally has a chance for top-notch ground games.

Frankly, right now, state level organization in places like Illinois and Ohio falls well short of the mark. Outstanding national level organizing and strategizing is going far better and Sanders incredible social media strength paying huge dividends.

So there it is: based on deep 538 database work, I very accurately predicted the biggest upset in American polling history, bigger even than Gary Hart shocking Walter Mondale in New Hampshire in 1984. That happened because pollsters and dismissive poll aggregators and forecasters bought into their own bogus narratives. The mistake would be for Bernie people to do the same thing.

Again, beware the Ides of March. 300 pledged delegates is the very barest minimum to stay viable, 350 puts the campaign in great position to finally shut up the critics and go all the way to the Golden State.

Will Hillary Clinton Beat Bernie Sanders in Any County Outside Wayne, Michigan?

Previously, I dove a bit deeply into data from FiveThirtyEight’s Facebook Primary to suggest that, properly adjusted to USPD, it predicts a very significant upset for Bernie Sanders in Michigan today. In fact, there isn’t a single county outside Wayne County (Detroit and suburbs) where the model predicts a victory, or even a close contest, for Clinton. This includes Genessee (Flint)! The model, furthermore, predicts only a very slight edge for Clinton in Wayne County with suburban and non-black voters making up for an approximately an 80-20 preference for Clinton by African Americans in Detroit.

For now adjusting 538’s Facebook Primary to USPD has fairly accurately predicted the outcome between Clinton and Sanders at the county level 97% of the time^ for counties with more than 1000 voters.

Michigan Primary AP

^86% of the time the model predicted the right county level winner between Bernie Sanders and Hillary Clinton. An additional 11% of races were predicted by the model to be tight, but the model picked the wrong eventual winner.

Read More