Chapter 1: Polling data
During the time of the 2016 primary election, Hillary Clinton and her allies perpetuated a narrative that nominating Bernie Sanders would translate into a Republican win in the general election. As it increasingly looked like Donald Trump would become the Republican nominee, team Clinton could then invoke the fear of Trump on the Democratic electorate, claiming that a Bernie Sanders nomination means a Donald Trump presidency.
During the primary election, though, it became increasingly clear that the opposite was true. The existing data favored Bernie Sanders, not Hillary Clinton, in a general election: in general election polls, Sanders fared better overall against the potential Republican nominees, and did particularly well against Trump.
The chapters that follow will explain why this is the case — why Sanders would have been so strong against Trump, where Clinton was not. But first, let’s look at the data that was available during the 2016 primary election.
In mid December, a month and a half before primary voting would begin, Rob Hager published an article on HuffingtonPost, “Who’s spoiling now?: Polling indicates that Democrats underrate Sanders’ electability at their own peril.”
In December 2015, in national primary polls, Sanders was losing badly to Hillary Clinton. But when pollsters asked a different question — the question of which candidate would be preferred in a general election — they found a very different result. Already, in late 2015, Sanders fared better than Clinton in general election polls:
Even in December 2015, when Sanders was way behind Clinton in national Democratic primary polls, more voters were saying they would vote for Sanders in a general election against a Republican candidate.
As illustrated in a previous report, one tactic used against Sanders in the 2016 primary was to cast him as “unelectable” in the general election, and boost Clinton’s image as the “inevitable” candidate. By collaborating with friendly mainstream journalists, and using other tactics, the public would continue to hear the “Sanders is unelectable” narrative, despite these polls that already in December 2015 were beginning to show otherwise.
There is another reason beyond “perception of electability” that Sanders was already beginning to poll well for the general election. The electorate for a general election is not the same as the electorate for a Democratic primary election.
From December 2015 onward, the margin would only grow. By March 2016 — (the primary election took place from February through June) — polls were consistently showing Sanders as the better candidate versus a Republican in a general election. And Sanders was polling particularly well — far better than Hillary Clinton — against the likely Republican nominee, Donald Trump.
In late March, polls would begin to show Sanders beating Trump by huge margins.
Some states were showing razor-thin margins for a “Clinton vs Trump” race, where Sanders was winning handily. (Later in the chapter we will look at this data more closely.) In March 2016, many in the Sanders camp were already getting worried that Hillary Clinton could lose to Donald Trump in a general election:
Just a few months into the primary election, never again would Hillary Clinton poll as high as Sanders did against Trump.
There were a lot of polls conducted, but the “average” polling data is most important to track. Near the end of the Democratic primary, on average, and with no exceptions, Sanders was polling better than Clinton against Trump. Zero polls showed Trump beating Sanders. Meanwhile, Hillary Clinton was struggling to beat Trump in some polls, and occasionally, Clinton was losing to Trump in a poll:
Here is a long-term look at Clinton’s average polling versus Donald Trump. Notice how often Trump comes close to Clinton, and for a few days in late July, Trump was polling better than Clinton — not just in a few polls, but in average polling.
Now compare the previous graph to “Trump vs Sanders.”
(Aside: If you look at the RCP polling average, you could also see that John Kasich had a solid lead on Clinton, while Sanders was beating Kasich.)
Bernie Sanders and his supporters were pointing to these polls all throughout the primary elections — here was the poll data showing that Sanders fared much better against the Republicans, and Trump in particular. But these concerns were dismissed by the media, the Clinton campaign, and the Democratic Party.
Let’s look at the polling-average charts again, side-by-side.
(Clinton left, Sanders right)
Near the end of the 2016 Democratic primary, polls of Clinton vs. Trump showed smaller margins for Clinton, with Trump sometimes winning:
In the same time-frame, Sanders had a much greater lead against Trump, and wasn’t losing in any polls:
One analysis by 270towin predicted a blowout election if Sanders faced Trump. Here is the electoral map:
The Real Clear Politics prediction was not quite as generous, but still very favorable to Sanders. Bernie Sanders only would have needed to win two swing states (that Clinton eventually lost) to secure a victory:
Then let’s look at some of those important swing states — that Clinton lost — and see if Sanders would have won them.
Clinton ultimately lost Wisconsin, which was somewhat of a surprise, as she was up ~6.5% in the polls around the time of the general election:
However, compare Clinton’s 6.5% lead to Sanders lead:
Sanders had an average +19% on Trump in Wisconsin, compared to Clinton’s 6.5%.
Is this “proof” Sanders would have won Wisconsin when Clinton did not? No, it is not absolute proof, but it is very strong evidence — it is proof enough to say that Sanders would have won Wisconsin.
There are some who would notice the difference in polling dates — Sanders’ last poll was in June, while the Clinton/Trump poll was in November — and would point to this as a reason that this is not good enough evidence. But, these folks would again be making the argument that Sanders would’ve dropped at least 15 points in Wisconsin — in about four months — to lose the state. That’s far more than Clinton dropped. Those better be some darn good “attacks” on Sanders. (Later, we’ll look at why these “attacks,” at best, would have done minimal damage to Sanders.)
Let’s look at Pennsylvania, another very important swing state. Around the time of the general election, Clinton was polling, on average, slightly ahead of Trump in PA — within the margin of error. That’s dangerous, because it means the election could pretty easily go in either direction. Pennsylvania eventually went to Trump.
Now let’s compare the available polling of Sanders vs Trump in PA:
Bernie Sanders was polling far better than Clinton against Trump in Pennsylvania. Again, is this “absolute” proof Sanders would have won PA? No, but it is a much stronger certainty than Hillary Clinton winning PA.
Hillary Clinton barely lost both PA and WI, and the final results were within a few points of the polling data. For Sanders not to have won Pennsylvania — just like Wisconsin — he would have had to lose a vast amount of support, in just a few months time.
Let’s look at another important swing state that Clinton lost.
Clinton was always set to lose to Trump in Ohio. But let’s look at how Sanders would have fared:
As you can see, Sanders was polling moderately ahead of Trump in Ohio. Would Sanders have won Ohio? We don’t know for sure, but the odds were with him, and certainly much better odds than Clinton winning Ohio.
Let’s look at a few more…
Clinton lost Michigan. Here was her polling data versus Trump:
Clinton was expected to win by a few points, but barely lost to Trump on election day. But let’s look at Sanders’ polls against Trump in Michigan:
These polls are far better for Sanders; much safer wins. We can accurately predict that Sanders would have won Michigan as well, an important swing state that Clinton lost.
Neither Michigan nor Pennsylvania have voted for a Republican president since 1988 (Source: Politico). That Hillary Clinton lost both these states was quite an upset, and a huge mistake. But there are significant, legitimate reasons why this happened (not just “sexism”), and some of these reasons will be explored in upcoming chapters.
Iowa is another swing state where Clinton was polling behind Trump, and ultimately lost.
But compare Clinton’s polling data to Sanders’ polling data in Iowa:
Sanders probably would have won Iowa, too, another swing state that Clinton lost.
There’s at least one more interesting swing state to look at: Indiana. During the time of the general election, Trump was polling far ahead of Clinton in Indiana — always greater than 10%.
Essentially, Hillary Clinton never had a chance in Indiana. And that’s fine — the Democratic candidate wouldn’t need to win Indiana, especially if they won some of the other swing states that leaned more “blue.”
And yet Sanders had a chance to win Indiana. First, like in Michigan, he won the Democratic primary in Indiana. Then, the one poll conducted of Sanders vs Trump in Indiana, in late April, showed Trump winning by only 1 point. This is within the margin of error, and could be overcome relatively easily (unlike a >10% margin, which Clinton was facing).
Surely the polls would have changed since that one poll of Sanders vs Trump in late April, but it is one poll that showed relative equality between Sanders and Trump — in a state where Clinton lost handily. It is also evident that Sanders’ popularity tends to go up as he becomes more familiar among the public (which will be re-visited throughout this report).
If we assume this one poll was relatively accurate — then in margins this close, it usually comes down to voter turnout on election day. As we will see in later chapters, Sanders was the candidate who inspired more “enthusiasm” among the broad electorate — not just the ones who could vote in the Democratic primary — and would have maximized voter turnout. If the Sanders campaign did just a decent job of getting out the vote in Indiana, Sanders probably would have won Indiana as well.
The polling data shows that from December 2015 onward, Bernie Sanders polled better than Clinton in a general election versus the Republican candidates. Sanders especially polled better against Donald Trump.
Clinton struggled against Trump in some polls, and for a short time (before the “Comey letter,”) was below Trump in average polling.
All arguments that Bernie Sanders would have lost to Donald Trump rest on a wild assumption that somehow, Bernie Sanders would have fallen much farther than Clinton did — in just a few months’ time, between July and November.
Why this is incredibly unlikely will continue to be explored in the report. Before moving on to the conditions and circumstances, though, there is one more set of data we must examine, which helps to explain why the general election polling data favored Sanders over Clinton.