Monthly Archives: March 2016

wisconsin democratic primary predictions

Wisconsin Democratic Primary Predictions

We’re back with our Democratic prediction model, which fared very well during Western Saturday (it correctly predicted the winner in each of the Alaska, Hawaii, and Washington caucuses and its vote share estimates also fell close to the actual results).  While those results likely did not change the trajectory of the race, they have certainly infused Bernie Sanders with momentum: In the past week, Wisconsin polls flipped from having Clinton up 6 points to Sanders being up an average of 5 points.

Our Wisconsin Democratic primary predictions show two different (and simultaneously expected) results.  The table below depicts win probabilities for the two candidates.  It largely aligns and mimics the polls — Sanders has a clear advantage and is indubitably favored, but not overwhelmingly so (a win probability one would expect with a candidate leading the polls by just more than the margin of error).

Hillary Clinton Win Probability

Bernie Sanders Win Probability


However, the vote share model tells a different story.  The vote share Wisconsin Democratic primary predictions point to a decisive, landslide victory for Sanders.  Our vote share model relies heavily on demographics and those of Wisconsin trend favorably to Sanders — the state is overwhelmingly white (82 percent) with a very small African American and Hispanic population (6 and 5.6 percent, respectively).  These demographics are similar to those of Minnesota, a neighboring state which Sanders handily won (with 62 percent of the vote; Minnesota also favored Sanders because it was a caucus).  Sanders fares very well with white voters and their large presence in the state’s electorate leads to the model advantaging him in the primary.  In other words, if he’s to make up the delegate gap, Wisconsin is very favorable terrain to net a large number of them.

Hillary Clinton Vote Share

Bernie Sanders Vote Share


Will our predictions bear out?  Based on polls, it seems so, though given Sanders’ recent momentum and financial resources (which could fund a substantial last-minute ad blitz), it would not be surprising to see Sanders win by slightly larger margins.  Considering that Wisconsin is 82 percent white, the predicted margin is actually rather disappointing for Sanders – favorable demographics in a medium sized state offer him an increasingly rare opportunity to pick up a large amount of delegates and begin to meaningfully close his deficit.  We predict the below delegate allocation:

Hillary Clinton Delegate Expectation

Bernie Sanders Delegate Expectation


These targets, again, seem reasonable given the polls.  If Sanders earns more than 46 delegates from the primary, it will be a good day for him.  If he passes 50, it will be a very good day for Sanders (though, unless indicative of beating polls and expectations, the single victory here will not alter any race dynamics).

As always, take these numbers with grains of salt as they reflecting underlying electoral conditions, not the campaigns or the candidates or momentum or news, etc.  These estimates may well be wrong (we fully admit that) and in the case they are, we’ll go right back to the drawing board to refine and edit our models.  Any comments about these forecasts or our models are welcomed!

[Be sure to follow us on Twitter and like us on Facebook for election news and analysis!]

alaska caucus hawaii caucus washington caucus delegates bernie sanders

Alaska, Hawaii, and Washington Caucus Recap

Bernie Sanders had a big weekend, notching victories in the Alaskan, Hawaiian, and Washingtonian caucuses.  He won each state by large margins – the Alaska caucus by 63, the Hawaii caucus by 69, and the Washington caucus by 45.  A boon to his momentum and fundraising (he brought in around $4 million in the days following the caucuses, ensuring he has the financial resources to invest heavily in Wisconsin and other upcoming primaries), the victories did not change the race’s trajectory, as we have documented.  That said, while it remains to be seen whether Sanders can turn these victories into lasting momentum, a positive sign for his campaign emerged from the night: Sanders beat expectations, necessary in all the upcoming caucuses and primaries if he hopes to win the nomination.

First of all, we define “expectations” as our delegate estimates based on PoliticalEdu’s model.  Democratic delegates are awarded proportionally, so we simply multiply vote share estimates by state delegate totals to create a benchmark to which we set our expectations.  The model explains around 96 percent of the variation between actual vote shares, so it is quite accurate and continues to “learn” as more results become available (we add each result to the model to hone and refine the relationship between variables, allowing the model to adjust to new electoral inroads a candidate makes and the changing primary landscape).  It is, of course, not perfect and thus our delegate targets are not perfect, but the model’s high degree of accuracy and focus on underlying electoral factors provides a strong baseline estimate; beating the model’s expectations shows a campaign that overcomes earlier trends and gains votes at the expense of the opponent.

With expectations defined, let’s turn to Saturday’s results.  Our model showed Sanders taking 74 percent of the vote in the Alaska caucus, sufficient for 12 delegates.  Here Sanders exceeded expectations.  He took 13 delegates from the Alaska caucus with 81.6 percent of the vote.  Though he clearly beat the model and its delegates expectations, doing so only led to his gaining a single extra delegate.  Alaska is an incredibly small state and only had 16 delegates at stake.  Beating expectations in Alaska is fairly meaningless – Sanders surpassed his vote share target by around 10 percent and got only one more delegate.  A remarkable feat lost significance because Alaska had so few delegates in the first place.  Sanders’ large Alaska caucus win netted him just 10 delegates over Clinton; he needs that margin of victory in large states where he can eat into the still very large delegate gap separating him from Clinton.

In the Hawaii caucus, our model again showed Sanders earning 74 percent of the vote and with it, 19 delegates.  Here he underperformed, getting (“just”) 70 percent of the vote and 17 delegates.  He fell two delegates off his target: already, his expectation-defying performance in Alaska is overshadowed by coming up short in Hawaii (up 1 delegate in Alaska but down 2 Hawaii, netting to being down 1 in delegate expectations).  As delegate number increases, Sanders simply cannot afford to underperform expectations, otherwise he will have little chance to close Clinton’s delegate lead.  Each delegate matters, especially with Sanders trailing by a substantial amount, and failing to meet expectations in any state – even a small one like Hawaii – makes his climb to the nomination a little bit harder.

Turning now to the Washington caucus, we forecasted Sanders taking 68 percent of the vote and 69 delegates.  He easily defied expectations by earning a staggering 73 percent of the vote and 74 delegates, surpassing expectations and securing his largest single-state delegate haul of the primary season.  Sanders was hugely favored in the state, which has small a African American subset and tends to be very liberal.  That said, once again beating vote share estimates by almost 10 percent is an impressive feat, especially considering we already had high estimates and expectations for his showing.  Importantly, unlike the Alaska caucus, the Washington caucus had a substantial number of delegates.  Beating his expectations by around 10 percent earned him an extra 5 delegates (as compared to the single extra delegate he picked up in the Alaska caucus).  Five delegates alone do not, of course, close the delegate gap, but Western Saturday showed that Sanders is able to surpass high expectations and drive up the score in states he should win.  Moreover, the Washington caucus showed that Sanders’ appeal is not limited to small states – a trouble that has so far plagued the Sanders campaign.  The question remains, though: can he beat expectations in large, diverse states – states through which the path to the nomination weaves?


Sanders Delegate Expectation

Actual Delegates



(A strong showing, but Sanders must make expectation surpassing a trend that carries over into the New York, Pennsylvania, New Jersey, and California primaries).

It remains to be seen whether these victories will propel Sanders in the upcoming contests.  He still faces a daunting task in closing Clinton’s lead.  Doing so seems highly unlikely, but if Western Saturday serves as precedent or starts a trend, Sanders may well overperform estimates in future primaries and caucuses.  Well it be enough?  Probably not – he’s trailing large in must-win states.  But habitually overperforming delegate estimates will give Sanders and his supporters hope and may result in tighter contests than many are expecting.

can bernie sanders win the democratic nomination

Bernie Sanders Swept “Western Saturday” – Here’s Why It Doesn’t Change the Race

Bernie Sanders lodged large victories in Alaska, Hawaii, and Washington on March 26. With his victories come a torrent of new pledged delegates, closing Hillary Clinton’s lead and refusing his campaign with a shot of momentum. But do his wins change the delegate math; can Bernie Sanders win the nomination?

To the first question, no; to the second, it still remains unlikely.

Alaska and Washington – two caucus states with low African American populations (with whom Clinton fares well) – represent states Sanders should have handily won (and, to be fair, he exceeded expectations). Washington, furthermore, is very liberal: Seattle is the third most liberal city in America and as the predominant (pro-Sanders) force in the state, helped Sanders rack up a large margin of victory. Sanders also spent substantial time in the state knowing that if his campaign had even the slimmest hope of notching 2383 delegates, Washington was a must-win.  Its demographics also were favorable to Sanders – a high white population coupled with a small African American subset played into Sanders’ strengths.  There’s a reason 538’s delegate target model showed a high Sanders target. Sanders did what he needed to do (and then some).

On the other hand, winning Alaska is fairly meaningless, as expressed by the below tweet from Washington Post reporter Philip Bump:
alaska caucus bernie sanders
The state has just 16 delegates at stake and even though Sanders overwhelmingly won the state, he netted few delegates over Clinton. A strong victory in a small state is more than offset by Clinton’s strength in large, diverse primaries – winning numerous Alaskas does not help Sanders win the nomination.  And once more, demographics favored Sanders: though home to many minorities, few African Americans (Clinton’s best demographic) reside in the state.  Alaska’s average age is also fairly young, another boon to Sanders.  He won a favorable state, hardly a reason to believe the race has flipped, though, to his credit, Sanders surpassed expectations.

Hawaii was a wildcard. Its unique demographics made the state difficult to predict, though our model still favored Sanders. Again, Hawaii is very small with only 25 delegates up for grabs. His strong showing here, as in Alaska, netted Sanders a small amount of delegates, far from the amount needed to overcome his 300 delegate hole (which does not take into account his superdelegate deficit).  The Alaskan paragraph more or less translates to Hawaii.

Winning small states, as Sanders has easily done, provides a morale boost but not a game change. The Democrats allocate delegates on a proportional basis, so even a huge victory in a small state results in single or low double digit delegate wins. Sanders has yet been able to expand his small state victories to large, diverse electorates. Those states hold large number of delegates and through those states the path to the nomination weaves. Yes, Saturday closes Sanders’ delegate gap, but he comes out of these contests still trailing by around 250 delegates. That can only be scaled with large-state victories.

Sanders needs to break serve; he’s winning the states in which he’s favored (beating expectations in those states), but these states are small and delegate poor. He needs to pull upsets in large, diverse states, the likes of which have eluded thus far. His Michigan upset needs to happen again to keep Sanders competitive.

So, can Bernie Sanders win the nomination?  Not unless he can translate these victories to momentum and poll movement, winning the Alaska, Hawaii, and Washington caucuses amounts to kicking a field goal when trailing by 18 in the fourth quarter of the Super Bowl.

democratic primary predictions and forecasts

Democratic Model

If you’ve been on our site, you’ve likely seen our Democratic primary predictions.  This post explains our models and how we come up with our Democratic forecasts.

Our first model predicts state winners; it uses a logistic regression and runs simulations for each election 1000 times.

The state winner model uses the following independent variables: whether the contest is a caucus, state percent white, black, and Hispanic, state GDP per capita, and the state’s average age.  The dependent variable is a Clinton victory (a logistic model with a Clinton victory coded “1”).  Here are its predictions for the primary season as well as the each primary/caucus’s actual winner:


Hillary Clinton Win Probability

Bernie Sanders Win Probability

Actual Winner


For Democratic primary vote share predictions, I use a dummy variable for caucus, state percent African American, white, and Hispanic, state GDP per capita, percent of the state population between 18 and 25, a dummy variable for whether Clinton is predicted to win (found above), and a dummy variable for the South.  The dependent variable is her actual vote share.  This approach explains around 93 percent of vote share variation.  Its predictions as well as actual outcomes are shown below:


Hillary Clinton Vote Share

Actual Clinton Vote Share


Each contest yields new data points that are then put into the model.  The two tables shown depict initial predictions, not outputs from after the model has considered new information and learned from it (which, when backtested, presents more accurate results than our first estimates – no surprise there).  Our Democratic primary predictions become more accurate as n (completed primaries/caucuses) increases, leading to a higher degree of confidence with which we can forecast the primaries.

Feel free to comment with questions or suggestions – we’re always looking to improve our model!

ted cruz affair ted cruz sex scandal

Donald Trump and the Ted Cruz “Sex Scandal”

Before the Utah caucus, the super PAC “Make America Awesome” ran a distasteful online ad (shown below) featuring a naked Melania Trump (Donald’s third wife) on a fur blanket.  The ad presumably attempted to appeal to Mormon values ahead of the Utah caucus, which Ted Cruz ultimately won with over 50 percent of the vote.  Trump rightfully took insult to his wife’s portrayal. ted cruz affair

In response, Trump sent the below tweet,  riddled with inaccuracies and uncalled for mudslinging.

ted cruz sex scandal

Here’s the problem: Make America Awesome PAC is unaffiliated with Ted Cruz.  The PAC is anti-Trump (not pro-Cruz or pro-Kasich).  Furthermore, and most importantly, it is against federal law for a candidate to communicate with a super PAC during the campaign.  Cruz therefore had no idea the ad was being produced or put out (regardless, he disavowed and condemned the ad later).  He could not tell them to make the ad, he could not tell them to run the ad, he could not tell them to cease the ad.  Any of those actions would have violated federal law.

But Trump kept propounding the inaccuracies.

ted cruz heidi cruz affair

This is wrong on all levels.  Cruz had no way of knowing the ad was being produced – otherwise he’d be violating federal law, something no candidate seriously running for office would do.  Trump’s tweet is wholly false and a baseless accusation designed for the sole political purpose of attacking Cruz.  There are no included truths.  Trump later continued the attack, retweeting this photo of Heidi Cruz, Ted’s wife:

ted cruz affair national enquirer

Disgusting and revolting and utterly below the dignity associated with an individual running for President of the United States.  Cruz responded by telling Trump, in no uncertain terms, to “leave Heidi the hell alone” (it is worth noting that Cruz has said nothing but nice things about Melania, effusively praising and complementing her.  Trump has extended no such courtesies).  Trump took an attack by an unaffiliated organization and used it as an excuse to blast Cruz’s wife without merit and without reason.

Enter the National Enquirer.  On March 25, the tabloid ran a damning piece alleging a Ted Cruz affair with multiple women.  Cruz vehemently denied the (likely fabricated) story and ripped Trump for planting the hack.  Trump denied all wrongdoing, though his statement lent backhand credibility to the story, fanning the flames of a Cruz sex scandal and throwing further his campaign’s muck.  The Enquirer has endorsed Trump and called for his election.  While it cannot be said with certainty that Trump planted the story, the connections are there and the timing is more than suspicious.

Regardless, Trump has stooped to new lows in shamelessly attacking Cruz’s wife.  The attacks stemmed not from a Cruz campaign ad, but from Trump’s seeming desire to drag Cruz into the muck, destroy his character, and senselessly slander his opponent.  These gross actions come from a presidential frontrunner and have no foundation in truth.

Trump must back down and apologize to Cruz and Heidi immediately for both his insults and his hinting of a Cruz affair.

[Be sure to follow us on Twitter and like us on Facebook for election news and analysis!]

washington democratic caucus

Democratic Forecasts: Alaska, Hawaii, and Washington

Our model correctly predicted each of the 3/22 contests and is now 28/29 for the “season.”  Below we show our forecasts for the Democratic caucuses in Alaska, Hawaii, and Washington:


Hillary Clinton Win Probability

Bernie Sanders Win Probability


Below is our prediction for caucus vote shares.  The model explains 96 percent of the variation between vote shares, so it is not perfect, but still highly predictive.  These numbers will not be perfect and will likely miss actual shares by 2-5 percentage points.  However, they provide strong vote share estimates and yield important delegate targets for each state.


Hillary Clinton Vote Share

Bernie Sanders Vote Share


Which leads us to delegate targets for the day:


Clinton Delegate Target

Sanders Delegate Target





Clinton should hope to win 42 delegates and Sanders 100.  Any numbers above/below these estimates would prove to be a good/bad day for the candidates.  Sanders is down around 300 pledged delegates – netting 58 over Clinton would close that gap and bring him some needed momentum as the map moves to his stronger states.  (Of course, netting 58 delegates does not put him in the lead and does not climb him out of his delegate hole – it simply moves him ever so slightly closer to Clinton.  Sanders needs to perform well continuously and win large states, not just small ones like Idaho and Utah).

Saturday looks to be strong for Sanders.  We’ll see how the day unfolds.

arizona primary voter suppression

3/22 Democratic Primary Predictions

Using PoliticalEdu’s Democrat primary model – which relies heavily on state demographics and is 25/26 so far – we have our predictions for the March 22 contests in Arizona, Idaho, and Utah.  Keep in mind these show the win probability, not forecasted vote shares (and so little information regarding delegates can be gleaned from this model; a vote share model is in the works).


Hillary Clinton Win Probability

Bernie Sanders Win Probability