This will be a relatively straight-forward article, largely because its central premise is inarguable: the recent declaration by the national media that polls and votes are the same thing means it’s entirely possible that Bernie Sanders is winning the Democratic primary. And since the extent of that probability can’t be discerned — given the media’s odd theory about polling and votes — as a Sanders supporter I think it’s fair for me to say that Sanders is definitelyleading the Democratic primary, at least as far as the pledged delegate count is concerned.
This assertion will undoubtedly confuse, bemuse, and enrage many people — so you’ll have to read on to understand it. Or, if you prefer, stop reading here and go ballistic about this article on Twitter and Facebook. It certainly won’t change the fact that, now that the national media has confirmed that polls and votes are the same deal, it’s perfectly likely that Bernie Sanders is winning the Democratic primary.
For those still reading, here’s what you need to know.
Nate Silver yesterday revealed that, on average, Democratic primary polling 21 days or less before the 2016 primaries and caucuses has been off by 10.6%.
That’s not a typo: 10.6%.
We can therefore conclude that the 54% to 46% lead that Hillary Clinton has over Bernie Sanders in pledged delegates is in fact a statistical dead heat, with a possibility that Sanders is actually ahead of Clinton in pledged delegates.
Wait a minute, you might say — that’s wrong, isn’t it?
And the answer is: it depends.
The only way to refute this assertion — the assertion that Bernie Sanders is currently ahead in the pledged delegate count — would be to say that “polls” are not “votes.” In other words, your wholly reasonable objection might opine, while polls can wrongly predict voting results, actual pledged-delegate votesdon’t have an MoE (margin of error) because they’re factual and constant.
By the same token, the media taking a poll of all Democratic superdelegates on June 6th, and then using those poll results to declare that Hillary Clinton has (variously) “won,” “clinched,” “secured,” or “gained” the Democratic presidential nomination, is nonsensical — as, like any poll, a poll of how superdelegates feel on June 6th, especially a poll taken in advance of those superdelegates knowing how Californians have voted, has an MoE.
The further out a poll is taken from the actual vote it relates to — in the case of the media’s phone-banking of superdelegates, well over a month and a half, as superdelegates don’t cast votes until July 25th in Philly — the higher the MoE.
What we don’t know is what the margin of error for the media’s June 6th superdelegate poll will be. The only MoE data we have on Democratic polling in 2016 is Nate Silver’s 10.6% figure — suggesting that, on average, primary polling in 2016 has been off by 10.6%. In other words, media pollsters have over and over again called Democrats who haven’t voted yet, tallied up their declarations of who they’re going to vote for, and then promptly issued a poll which turned out to be, on average, wrong by around 11%. That’s quite a lot.
In normal polling, margins of error are introduced by sampling — pollsters’ inability to query the entirety of a population, which requires them, instead, to estimate the whole based on a representative sub-set. In superdelegatepolling, an entire population is polled, but (a) a significant percentage are in effect non-responders, as they won’t declare their preference to pollsters, (b) error is introduced by the fact that the population is being polled significantly in advance of their late-July balloting, and (c) the polling occurs in a context in which the media itself has publicly said that it expects there will be much movement in the opinion of the polling population between now and late July. For these reasons we can be certain there will be error — a margin of error — in the media’s superdelegate poll, just as there is with any other form of poll. The same cannot be said for actual vote tallies, which is why we reasonably expect them, and only them, to be used to call national elections in America.
Ultimately, the same unpredictability that’s infused this entire election cycle and produced an astronomical MoE in conventional polling can and likely will do the very same in the media’s polling of superdelegates. In just the last two weeks, for instance, Sanders has gained a handful of superdelegates, despite the national news media beginning to call a Clinton win inevitable weeks ago. And for those who say that, despite new announcements from superdelegates, there is little evidence of any already-polled superdelegate switching up their now-temporary endorsement, we can note the following: the very reason no superdelegate (bar one) has flipped from Clinton to Sanders or Sanders to Clinton is, ironically, the very reason the polling itself can’t be relied upon: (a) the vote is too far off for anyone to need to announce a change of heart, and (b) nearly all of the events that would change minds — for instance, the vote in California or the conclusion of an FBI investigation — haven’t happened yet.
So factors that could affect the margins of error for the media’s June 6th and subsequent superdelegate polls include: Sanders winning California; Clinton being indicted or otherwise scarred by her still-pending FBI investigation; Joe Biden entering the race a few days or weeks before the DNC Convention as part of a Biden/Warren unity ticket; or any one of a million other news events that could somehow swing superdelegates against Hillary Clinton. What if her unfavorability rating finally inches up to 70%? What if she goes the next two months without leading Donald Trump in even a single national poll? What if her Wall Street transcripts finally get released, revealing that nothing she’s said about her economic policy positions and first principles is true? Are these far-fetched possibilities? Actually, given what’s already happened, not really.
What if Clinton made some campaign-ending gaffe, such as accidentally being caught on-camera or on-mic saying she doesn’t give a damn about the youth vote, or the labor vote, or the white vote? And on and on and on — in this most unpredictable, as in 10.6% unpredictable, of election cycles, who the heck knows what could happen? Donald Trump could debate Bernie Sanders in D.C. on June 11th and lose so badly to Sanders that millions of Americans write on social media that they want to switch their earlier Clinton votes to Sanders votes. Stranger things have happened, and will continue to happen.
The point: polls aren’t equivalent to votes because they have margins of error.
And if you believe, like so many Clinton supporters and media analysts do, that polls are equivalent to votes, then the margin of error for Hillary Clinton’s 54%-46% pledged-delegate lead over Bernie Sanders is, at least normatively, 10.6% —meaning that Sanders may actually be ahead by as many as 2.6% in the pledged delegate count. If the pledged delegate vote — or let’s say “poll,” as again, per the media, what’s the difference, really? — is really inaccurate, Sanders might be up by as many as 5 or 6 points on Clinton, in which case she has little chance of leading in pledged delegates even if she performs very well on June 7th.
But wait! It gets better.
A few weeks ago, John King of CNN told his viewers that if Sanders performed exceedingly well in Indiana, West Virginia, Oregon, and Kentucky — which he did, winning the first three and drawing in the fourth — and then ran the table or nearly so on June 7th, “many” superdelegates would switch their allegiance and the media would capture that sea change and report on it.
Except King was lying.
We know this because the major networks have recently announced that they will “call” the Democratic primary for Clinton — declare her its “winner,” say she’s clinched the nomination, or whatever parlance equates to those terms — without first polling the superdelegates post-California.
In fact, they’ll release their poll hours before California has tallied any votes.
So on June 7th Bernie Sanders could win every state, or even just California and three others, a circumstance King said would undoubtedly flip “many” superdelegates’ votes, and America would never find out what the margin of error of the media’s June 6th superdelegate poll was because Clinton would have “won” the nomination before that margin of error could be determined.
Needless to say, even were the media to re-poll the Democratic superdelegates on June 8th, and to report on June 8th whether their own June 7th call of the primary for Clinton was accurate, that re-polling would immediately develop its own margin of error the moment any intervening event occurred that could affect even a single superdelegate’s view of Hillary Clinton.
This may be why the Democratic National Committee told the national media,clearly and repeatedly, not to poll superdelegates prior to the DNC Convention.
But the media did it anyway — indeed, they did it with the blessing of analysts like Nate Silver, Nate Cohn, and Harry Enten who knew how high the MoE was for Democratic primary polls this year. And yet suddenly these analysts’ fears of a 10.6% MoE in primary polling — or perhaps worse than that! — dissipated at the very moment getting the facts right mattered the most. Maybe it’s because their very jobs depend on having something to do and write about at this phase of the primary season? So maybe it’s not actual news running the show now, but the realities of media corporations’ bottom lines and payrolls?
It’s like saying, “Don’t trust that poll of Alabama voters that was taken 3 days before the primary there, it’s probably way off,” but simultaneously saying, “accept as the veritable Gospel a poll of a different group of Democratic voters taken 49 days before they vote” — the latter poll being Gospel because, in the view of the media, it determines who “won,” “clinched,” “secured,” “gained,” “has” (or whatever word you like) the Democratic nomination for President. Read any article you like from the past two weeks, or in the next few days — whether it’s in The Washington Post, The New York Times, or anywhere else — and you’ll find phrases such as “clinched the nomination,” “secured the nomination,” and “won the nomination,” with nary a “presumptive” in sight.
So what’s the difference between an Alabama Democrat and a Democratic superdelegate?
Simple: the former might change their mind before they vote, because they’re open to the influence of intervening events that happen between the time they’re polled and the time they cast a ballot.
Meanwhile, because the superdelegate system is rigged to favor pre-selected Establishment candidates, the media is certain the MoE of their superdelegate polling 49 days before the Democratic National Convention is virtually nil.
And no, it’s not because Clinton leads in popular votes and pledged delegates — as in fact superdelegates weren’t created to rubber-stamp the popular vote or pledged-delegate tally, this year’s Democratic superdelegates have explicitlyand en masse told the media that that’s not their role, and even Hillary Clinton herself said in 2008 that that’s not what superdelegates are or do.
To repeat: the media is conceding that the superdelegate system is rigged by refusing to acknowledge that their polling of superdelegates has an MoE — and that therefore Clinton can’t be said to have “secured” the Democratic nomination at any time prior to the vote on that question that will take place on July 25th in Philadelphia.
Shorter: If superdelegate polls have no MoE, the superdelegate vote is rigged.
And if they do have an MoE, they’ve no business being used to call an election.
Now, I’ve been accused of using “experimental math” in my columns — accused by New York Magazine, accused by The New York Times, accused by others within and without the media — so let’s nail this thing down:
- “Experimental math” happened every single time the mainstream media reported that Bernie Sanders having to win 67% of pledged delegates on June 7th to take the pledged delegate lead was “mathematically impossible,” an “insurmountable” task, a “foreclosed” possibility, or any of the other experimental-mathematical terms mainstream media outlets like the ones listed above used to turn 67% of votes on a single day of voting into “infinity”;
- “experimental math” happened every single time the mainstream media added a superdelegate “tally” — a poll — to their actual on-screen votingresults, even though the two types of data cannot be combined in part because the former has an unreported and indeed (in this instance) unknowable margin of error;
- “experimental math” will happen on June 7th when the news media declares Hillary Clinton has “won,” “clinched,” or “secured” more than 2,383 delegates, when in fact she will have won, clinched, or secured 500+ delegates less than whatever that final tally is, at least if the words “won,” “clinched,” “secured,” or “unpledged” continue to have a meaning;
- “experimental math” happened each and every time one of the horde of horse race-obsessed number-crunchers who have now replaced journalists in the news media blithely equated two contextually and historically distinct events — say, the Washington caucus in which votes that mattered were tallied, and the Washington primary in which all who participated knew that nothing they said or did mattered one whit — using the magic of pie charts, graphs, and alleged mathematics;
- “experimental math” is systematically under-analyzing and then under-reporting polling cross-tabs that reveal the divide between Sanders and Clinton supporters has always been one of age, not race or ethnicity, a fact we now see in high relief as Sanders competes with great success for nonwhite votes in a state (California) whose nonwhite voting population is disproportionately younger than that of other states;
- “experimental math” is systematically both under-recording and under-reporting differentials between early voting and live voting, as these reveal not only the fact that Clinton voters skew older (which we always knew) but also that her name recognition in the lead-up to a primary or caucus — and indeed in this entire primary season — accounts for an appreciable percentage of her popularity;
- “experimental math” is conducting data-driven analyses of this election without also reflexively polling and reporting on how the election has been covered by the media, for instance by taking a poll or running a feature on how Americans who get most of their news from TV coincidentally support Clinton, while those who get their information from the far more hybridic, heterogeneous, and dynamic news environment to be found online tend to support Sanders;
- “experimental math” happened every time exit poll results were adjusted after the fact based on nothing but their own failure to comport with the final vote tally — as indeed exit polls are always adjusted — thus creating what is at once an entirely artificial feedback loop and also, ostensibly, to hear the media tell it, an actually probative data-set; and
- “experimental math” is any condescendingly conclusory commentary on the “final results” of a primary or caucus (such as those in New York and Arizona) where provably and incontrovertibly thousands or even many tens of thousands of voters who woke up intending to vote were, by dint of actions not their own, kept from doing so.
What is not “experimental math” is noting how poorly Clinton has fared in the final two-thirds of the Democratic nominating process, or how poorly she fares in live voting, or how unpopular she is among the nation’s all-important independent-voter bloc, or how historic her unfavorables and “untrustworthy” ratings are, or how much worse than Sanders she fares when matched up against Donald Trump in either national or battleground-state polls, or how popular Sanders is among younger nonwhite voters, or how appalling it is that a candidate with every possible advantage politics can bestow on a human — universal name recognition; a proven political “brand”; the most super-PAC money ever compiled; a 400-superdelegate lead that dramatically influenced media reporting; endorsements from state officials with the ability to move votes; prior experience running a presidential campaign; and so on — couldn’t get, in a two-person race, the slight majority of pledged delegates (~59%) that would’ve been required to avoid recourse to superdelegates altogether.
Hillary was sixty-something points up on a rumpled 74 year-old independent socialist from a state where cows outnumber people and she blew almost the entirety of that lead — just as she’s blown (per MSNBC reporting) every leadshe’s had in every election she’s run in, as Hillary is not a person whose poll numbers have ever improved with additional campaigning or public exposure.
Don’t believe me? Look it up. Maybe some superdelegates finally will, too.
Nor is it “experimental math” to observe that Clinton is now in a dead heat with Donald Trump, with no hope — if history is precedent — of widening that lead unless many more Sanders voters come home than the 55% a recent poll by YouGov says will do so.
Nor is it “experimental math” to use the media’s own preposterous maxims to come to an equally preposterous conclusion about who’s winning the primary on the Democratic side.
And finally, it’s not “experimental math” to require a candidate to win an election only and ever when they’ve been declared to have done so by the only body with the authority to certify the results of the relevant vote: in this instance, the Democratic National Committee. And yes, that’s why we use the term “President-elect” the day after a national vote; we don’t say “President” because the candidate isn’t the President yet, but we use the term “President” in some fashion because the relevant bodies — the Secretary of State’s offices in fifty states; the political parties themselves; and the losing candidate — have conceded the election is over. Without either (a) a concession, or (b) a certification of a final vote, any suggestion by the media that a nominating contest is over is a violation of the most basic principles of journalistic practice. Reporters don’t ever get to call elections won, clinched, or secured.
What the media does is tell us when another entity has said an election is over.
As a university journalism professor I don’t care — nor do Sanders supporters; nor does the discipline of journalism, which once had an ethical code; nor does “math” — that the media has wrongly called primaries using phone polls in 2016 (on the Republican side) and every election before that dating back to 1984. Doing your job wrong on a Monday doesn’t explain why it’s all right to keep doing it wrong on Thursday. And whatever butt-hurt the media finally righting a wrong practice might cause Clinton supporters, the right thing to do within the discipline of journalism is ever the right thing to do — full stop.