Recently, many have called upon Bernie Sanders to drop out of the race for the Democratic nomination and for the party to unite behind Hillary Clinton in order to defeat Donald Trump. These people most commonly cite the math – saying that Bernie cannot catch up in pledged delegates – or the popular vote – noting that Hillary today has accrued more votes so she is the people’s choice. Hillary’s own website, in fact, makes the claim that she is the only candidate who can beat Trump because she’s the only one with more votes than him today (https://www.hillaryclinton.com/feed/trump-will-be-republican-nominee-only-one-candidate-can-stop-him/). The argument, then, is that Bernie will do more harm than good by staying in the race because he will wound Hillary in preparation for a general election fight against Trump.
Naturally, Bernie supporters did not take to this kindly, saying that if everyone who wanted to vote (e.g., Independent or unaffiliated voters) could actually vote in the Democratic primaries, then Bernie would be winning, and they took to Twitter with the #DropOutHillary hashtag, which went viral. This, in turn, prompted outrage from Hillary supporters who called Bernie supporters delusional and sore losers who need to do the math.
Name-calling aside, it seemed like the best way to settle the debate would be to actually do the math. My instincts told me that even if I included Independent and unaffiliated voters, that Bernie may not come out on top. However, when I did indeed do the math, I was wrong. It turns out Bernie Sanders is likely the strongest candidate today if Independents and unaffiliated voters could vote in all primaries.
Now before we get into the math, consider the following: there is no hard evidence today that shows that Hillary Clinton offers the American people a better chance to beat Donald Trump than Bernie Sanders – in fact, it’s the contrary. First of all, Bernie consistently beats Trump by a wider margin than Hillary beats Trump in every national poll. Secondly, Bernie has won over Independent and unaffiliated voters by a roughly 68%-27% margin on average in all open primaries and caucuses to date. And finally, Independent voters make up about 42% of the American electorate. This means appeal to Independents is absolutely crucial.
(Also before I continue – I understand that most Independents tend to lean one way or the other. Latest data indicates that 16% out of the 42% of America that is Independent leans Democrat, 14% lean Republican, and 12% are truly Independent. However, even if you exclude this 14% that will likely vote Republican, and split the 12% true Independents 50-50, then the 22% of the American electorate that is Dem-leaning Independent has to choose between Bernie and Hillary, but in many states they cannot vote.)
I’ve run the numbers state-by-state to see what the primary race would look like today if Independents and unaffiliated voters could have participated everywhere. In the scenario that accounts most closely for today’s data, Bernie would be trailing by just ~226k votes today, and would in fact be leading by ~50k votes by the end of the primary season, eking out a win in the popular vote (see screenshot of the model later in this piece).
This is YUGE as Bernie would say. Let’s go back to Hillary’s claim that she’s the only one who can beat Trump because she’s the only one who has more votes. This, now, is a stark misrepresentation of the true sentiment of the American people. The fact that so many people have been disenfranchised has basically altered the course of the entire Democratic primary and the dialogue around the nomination.
Given the fact that Bernie polls so much better than Hillary against Trump, given that he has wonderful favorability ratings (and in fact much better favorability ratings than Hillary), AND given that he appeals so much more broadly outside of the Democratic base, it is actually mind-boggling that people are asking him to drop out and unify the party behind Hillary. It is akin to having a young Michael Jordan on your team going into the fourth quarter of Game 7 of the NBA Finals, but choosing to bench him in favor of a better-known, established player just because sometimes they’ve won some games in the past against some opponents but you never really know how this person will play on a given day. I repeat, mind-boggling.
At this point, I’m sure Hillary supporters are clamoring that this has to be wrong, and there are many arguments to be made. So let me start by breaking down the mathematical approach here, and then address some arguments I’ve crowdsourced over the past week. My methodology was as follows:
(1) I went state by state based on the type of primary and looked at actual vote counts recorded, actual voter turnout % (e.g., how many Democrats turned out versus how many were registered), # of democrats and # of independents or unaffiliated voters in each state, and % of independents in the state that would likely be Dem-leaning (this comes out to 52% -- if you take the 16% of the 42% that are Dem-leaning plus half of the 12% true Independents, this gives 22 out of 42, or 52% of registered Independents in each state that could have participated if primaries were open). This yields the number of independents or unaffiliated voters that you'd expect to vote in the Democratic primary in each state where they weren't allowed to vote so far. Note that I have not even included an increased voter count in many caucus states, where it is hard to track individual votes.
(2) I built a "to-date" scenario, and also extrapolated to the remaining states based on expected turnout, etc. For the states that have already voted, we took the turnout that was actually observed in the state, and for the states that are remaining, we assumed that the voter turnout would be 30% (i.e., 30% of just registered voters, not 30% of the overall eligible-to-vote population). We also assumed that in remaining states, the vote would split 60-40 in favor of Hillary among registered Democrats and that the vote would split 68-27 in favor of Bernie among Independents and unaffiliated voters (the 60-40 split is a fairly accurate average of how registered Democrats have voted thus far, and an explanation of the 68-27 split for independents follows in the next paragraph).
(3) To get to the 68-27 average % of independents that would be expected to vote for Bernie versus Hillary, I looked at a combination of actual results in places where independents could vote (e.g., open primaries) plus latest polls from other locations both where Bernie won and lost. A couple examples of the independent vote split where Bernie won the state are Michigan (71-29 Bernie) and Wisconsin (62-31 Bernie), and a couple where he lost are Texas (57-41 Bernie) and Illinois (70-30 Bernie). Now, if you apply this methodology to New York alone, Bernie would have closed the gap by a net of 382k votes, and similarly Florida alone he would have closed the gap by 472k votes. If you now use this 68-27 average, account for voter turnout, and account for how many Independents lean Democrat, then like I mentioned, Bernie today would only be behind by ~226k in the popular vote, and by the end of the election season he would be up by ~50k votes.
This lays waste to both of the Hillary campaign claims that she’s the only one with more votes than Trump and that she’s the only one who can beat Trump. In fact, Bernie would absolutely have more votes than Trump if independents and unaffiliated voters could have voted everywhere, and he would also basically have the same number of votes as Hillary today! The split percentage-wise would have been 50.4%-49.6% in Hillary’s favor, and under our assumptions, by the end of the primary season, Bernie would eke out a win: it would be 49.9%-50.1% in Bernie’s favor (17.62 million votes for Bernie versus 17.58 million votes for Hillary).
It is important to note, however, that this story would change if (a) voter turnout were to drop, or (b) if the split of the Independent vote changed. For example in (a), if voter turnout dropped to just 25% in the remaining states, then while Bernie would still be ~226k votes behind today, he would only just barely pass up Hillary by the end of the season, winning the popular vote by ~2k votes (17.06 million votes for Bernie versus 17.05 million votes for Hillary). In scenario (b), we can keep voter turnout at 30% but simulate a much more favorable scenario for Hillary where she gets one-third of the Independent/unaffiliated vote rather than just 27%, so that the split is now 66.6%-33.3% in Bernie’s favor. In this scenario, Hillary would be leading by ~767k votes today and by the end of the season, she'd maintain her lead by about ~694k votes (18.19 million votes for Hillary to 17.49 million votes for Bernie, or 51.0%-49.0%). Taken as a whole, these scenarios show us the race is actually significantly closer than the recorded popular vote results might indicate, and that Bernie’s claim that he wins when voter turnout is higher holds true. If we chose to simulate results based on the parameters that most closely mirror today’s reality across all voters, Bernie would just barely come out on top.
Imagine how different the national dialogue (TV, newspaper, social media, etc.) would be if this was the state of the race today! Bernie supporters have been clamoring for months that the sentiment (and enthusiasm) of this nation was vastly different from what Democratic primary results show. Clearly we have to see how the final states play out, but based on the actual numbers and modeling, the country’s real sentiment is entirely different from the "presumptive" dialogue the news media has about Hillary.
Now, whether or not people believe in open primaries is an entirely different debate. But what this analysis shows is that today’s system of closed primaries is clearly not representative of how people in this country actually feel, and if we had a system that was more representative, Bernie might be winning the Democratic nomination on basis of the popular vote. The math here backs it up.
Take into account how much better his favorability ratings are than Hillary’s, and how much better he polls against Donald Trump, and it may even be safe to say that Bernie Sanders is likely the strongest Democratic candidate today. Long story short, if everyone had a voice, he’d likely be winning. Let me repeat from before: mind-boggling.
You can see the state-by-state rigor that went into the model in the table below – each of the actual vote counts, etc have been pulled from the state websites with election results.
You can also see the results of all three simulations in the table below:
Feedback (1): Hillary leads the popular vote as it stands today in our current system, so we have to abide by that. And furthermore, open primaries are not a good system because they allow people to sabotage votes by voting in the primary of the party they do not support just to help another candidate.
Answer (1): Let me take these one by one. First, it is true, Hillary does lead in today’s popular vote and I respect that sentiment. But I seems a travesty to me that all evidence points to one candidate having more support and being a better general election candidate, but that candidate may not be able to win in our system today. How good do we really feel about our democracy if results do not truly reflect the will of the people? If we are okay with the idea that someone can win a nomination in this country despite not actually having a plurality of support, then we can accept the conclusion that Hillary leading today means she should be the Democratic nominee.
Second, it is also true that people can cross over and try to sabotage the primary of the party they do not represent. But what evidence to we have to back up this idea? More important, what evidence do we have of its significance or magnitude? This is an entirely sentiment-based argument. Bernie supporters, given how ardent they are, are very unlikely to vote in another primary rather than voting in their own. And given Republican opposition to Trump and huge turnout in their own primaries, do we really think many of them would choose to vote in the Democratic primary instead and vote for Bernie to sabotage Hillary, or vice versa? The magnitude of the shift this would cause is minimal at best.
Feedback (2): Republicans have attacked Hillary aggressively, but have not yet done the same to Bernie. When they take off the kid gloves and actually do attack Bernie, his favorability and polling numbers will severely drop, and therefore, this idea that he is a stronger candidate or would receive more votes is false.
Answer (2): First of all, this argument is entirely sentiment based and has no hard evidence to back it up. One person even told me that Hillary went from being the person with the highest approval rating in the country to one of the lowest just because of attacks in this cycle, and therefore the same would happen to Bernie. The first part of this is patently false: on the day Hillary announced her candidacy on April 12, 2015, she had a 46.9% favorable versus 47.1% unfavorable rating – clearly not the best in the country. Furthermore, even before Trump announced his candidacy on June 16, 2015 – i.e., before he even started attacking her – her numbers had already slipped to 45.4% favorable versus 48.5% unfavorable. By the eve of the Iowa caucus, her numbers were at 42.1% favorable to 52.8% unfavorable, and today they stand at 42.4% favorable to 54.2% unfavorable. In the same time span, Bernie was 16.0% favorable to 16.7% unfavorable in April 2015, 22.1% favorable to 22.6% unfavorable in June when Trump joined, 47.3% favorable to 37.9% unfavorable on the eve of the Iowa caucuses, and 52.9% favorable to 39.6% unfavorable today. Bernie is, in actuality, the most popular member of the entire US Senate. (see http://elections.huffingtonpost.com/pollster/hillary-clinton-favorable-rating and http://elections.huffingtonpost.com/pollster/bernie-sanders-favorable-rating for the rating sources.)
But let’s put these numbers aside for a second. The only way this “kid gloves” argument even stands is if (a) you can directly relate Hillary’s dropping ratings to just GOP attacks, and (b) as a result of Republican attacks, more voters would defect from the Bernie camp than would from the Hillary camp. On point (a), Hillary’s drops in favorability are likely linked to tons of variables, including her email scandal, her campaign’s alleged use of push polls where they test out negative sentiments on voters and ask them if it would make them less likely to vote for Bernie, her eventual adoption of Bernie’s positions in her stump speeches, and her campaign not truly highlighting issues when she first announced candidacy (watch her announcement if you don’t believe me: https://www.youtube.com/watch?v=N708P-A45D0, her only mention of any substantive issue is around 1:40 where she says “the deck is stacked” – contrast this to Bernie’s candidacy announcement).
On point (b), break down the effects into branches. First, Republicans listening to Republican attacks are already highly opposed to Hillary, and would likely not vote for Bernie anyway. Second, Democrats listening to Republican attacks will unlikely be swayed to vote differently; those who are already committed to voting for Hillary still remain on her side, meaning they’d likely still vote for Bernie in the general election even if Republicans attacked him. And those already on the Bernie train aren’t going to switch in a significant manner. Finally, independents are already overwhelmingly in Bernie’s camp. Sure, he might lose some if Republicans attack him extensively, but a huge number of Republican-leaning independents won’t even vote for Bernie or Hillary in the first place. So the net effect of Republican attacks likely still leaves him on top versus Hillary.
Feedback (3): Independents have broken 68-27 for Bernie so far, but this can’t be used to approximate how things would’ve gone in other states since so many states have been closed contests, and further they cannot be used to predict the rest of the race.
Answer (3): True, you can never predict things with 100% certainty. But this 68-27 average also takes into account the latest polls even in closed primary states (e.g., Iowa was 69-26 Bernie, NY was anywhere from 80-20 Bernie to 70-30 Bernie). Again, this is not a perfect methodology for sure. But it seems to me the best we have. Like I’ve written above, you can model some other scenarios where Independents split differently, and this could lead to different results.
Feedback (4): Independents will not turn out to vote at as good a rate as partisans, especially because reports today show that Democratic turnout is not as high as it was in 2008 with Obama, for example.
Answer (4): Firstly, all evidence points to the idea that Independents are turning out at a significant clip – on average, data indicates that in states where Independents / unaffiliated voters can vote, the total # of votes is roughly 30-40% higher than if the votes were restricted to just Democrats. In fact, in the model here, the average increased voter turnout is a bit below 40%, but ranges from 20-50% depending on the state (~20% higher in Pennsylvania, ~33% higher in NY, ~50% higher in New Hampshire).
That being said, it’s absolutely true that a different voter turnout would lead to different results – the scenarios I wrote about above account for two of these possible scenarios, with either 30% or 25% of registered voters showing up to vote.
Feedback (5): You cannot assume that the vote will split 60-40 in all remaining states between Hillary and Bernie. In fact, Hillary might win more than 60% of remaining votes if current patterns hold.
Answer (5): True, it is very likely that the remaining split will not be exactly 60-40. But for now, that seems to be the best proxy given how the registered Democrat vote has split in many states (e.g., 57-42 in New York). Thus, this 60-40 split is actually the best representation of what would likely happen going forward. Furthermore, some upcoming contests seem to favor Bernie (e.g., West Virginia, Oregon) while others do not (e.g., Kentucky, Washington DC). California polling shows Hillary ahead now but this polling is largely outdated. Thus, the best we can do is use an average approximation for now.
Long story short, the methodology here is not perfect. But it gives credence to the idea that Bernie Sanders should indeed stay in the race and that he may actually be the choice of more of the American people than Hillary Clinton.