I know this is very long, but I always do my best to articulate point-by-point when refuting claims that are not based in reality. If you’ve seen this piece going around, then it pays for you to read it. Likewise, there are many talking points in here that I address that are just not grounded in truth; I do my best to bust them.
I suppose I'm just going to go into "disinformation busting" mode until this primary is over. Here's the latest piece that I've seen about 7 times in my FB feed in the past 24 hours. Written by Seth Abramson, “Bernie Sanders Is Currently Winning the Democratic Primary Race, and I’ll Prove It to You” argues that Sanders really is winning the Democratic primary — if we conveniently ignore most of the reality on the ground. He makes a few good points, but the conclusions he draws are mostly incorrect.
An Assistant Professor of English at University of New Hampshire, his area of expertise does not appear to be related to political science, psephology or any of the relevant topics at-hand. That’s not to say educated people cannot have opinions and make good points without such a background. However, many people citing HuffPo as of late don’t realize that it has become a more-glorified DailyKos in that virtually anyone can get published if you jump through a couple of hoops. I’ve been considering it myself!
Let's go point-by-point through this article.
"Bernie Sanders has terrible name recognition in states where he hasn’t advertised or campaigned yet; meanwhile, Hillary Clinton has universal name recognition everywhere. Realizing this, the Clinton camp pushed hard to rack up the early vote in every state where early voting was an option. They did this not primarily for the reason we’ve been told — because Clinton performs well among older voters, and older voters are more likely to vote early than other age demographics — but rather because they knew that early votes are almost always cast before the election season actually begins in a given state.”
PPP found in early February that 86% of registered voters nationally had an opinion of Sanders, compared to 95% for Clinton. If people do not recognize a name when polled, then they usually tend to not have an opinion of that person. Because of this, favorability numbers often serve as name recognition indicators when name recognition data is not available in polling. If 15% of voters have no opinion of a candidate, then it is safe to assume that the number of people who haven’t heard of the candidate is not larger than that.
This poll was among all registered voters: the likelihood of Democratic primary voters being less familiar with the candidates than the registered voting bloc at-large (many of whom aren’t following the primaries) is next to zero if you consider the common sense behind it. It’s likely that among that group, name recognition for both candidates is very close to 100% — and especially by now.
This also isn't 1930: news is nationalized, so is social media, and likely primary voters have been digesting opinions of the candidates from many different sources for months. A small boost in earned media from in-state visits can increase awareness, but again: we are talking about Democratic primary voters. It’s only a negligible percentage of likely primary voters in every state who have not already been following this race in a multi-faceted capacity.
He is right in saying that younger voters are less likely to vote early, but he is incorrect in suggesting that most people vote before election season has begun in a given state. Again, these races are heavily nationalized. Most states do not get the IA/NH/NV/SC treatment. People do not suddenly start paying attention upon the first visit or the first campaign commercial in a state when your state votes in March or later. Volunteers and field organizers even in these later states will have been on the ground for months in advance of primary day; thousands, hundreds of thousands and even (in some cases) millions of impressions will have already been made through these efforts.
Remember that the primary voting bloc and the general election voting bloc are not the same. A square is a rectangle but a rectangle is not a square. For every primary voter, there are 4 to 6 GE voters. Primary voters as a whole are much more involved in terms of consistently informing themselves about current political events. GE voters may wait until the final weeks to begin following campaigns, but primary voters as a whole begin following the candidates weeks, months or even years before that.
“That’s right — in each state, most of the early primary voting occurs before the candidates have aired any commercials or held any campaign events. For Bernie Sanders, this means that early voting happens, pretty much everywhere, before anyone knows who he is. Certainly, early voting occurs in each state before voters have developed a sufficient level of familiarity and comfort with Sanders to vote for him.”
We’ve already covered above how primary voters are not restricted nor usually refrain from learning about the candidates until one or two weeks before their states vote. We’ve also already covered how Bernie’s name recognition at-large is not an issue in these contests as of now. However, let’s briefly touch on the concept of campaign visits and commercials.
There are legitimate arguments as to whether campaign visits to a state — and campaign commercials, for that matter — in the modern era even have an effect. Most campaigns do still err on the side of caution, but there is no evidence to suggest that campaign commercials — especially positive campaign commercials — persuade voters. Campaign commercial spending in the 2012 presidential election totaled roughly $1,000 per truly undecided voter in battleground states (referenced in link above).
“But on Election Day — among voters who’ve been present and attentive for each candidate’s commercials, local news coverage, and live events — Sanders tends to tie or beat Clinton. In fact, that’s the real reason Sanders does well in caucuses. It’s not because caucuses “require a real time investment,” as the media likes to euphemistically say, but because caucuses require that you vote on Election Day rather than well before it.”
We’ve already covered how there is no evidence to suggest campaign commercials and live events have any bearing on voter behavior. Mr. Abramson points out that Sanders does better with Election Day voters than with early voters: this is largely true.
However, his claims that Sanders does better in caucuses because people have to vote on that day — as opposed to the more accepted claim that caucuses require more effort and therefore attract those most enthusiastic — cannot be proven. He could be right, or it could be the more commonly-cited reason. At the end of this piece, I address a huge part of why Sanders does better with ED voters, which is a point that he acknowledged at the beginning of his piece (but did not recognize as being the cause).
EDIT: I WAS PROVIDED IN THE COMMENTS BELOW WITH A LINK TO A WP ARTICLE BY PHILIP BUMP THAT ALSO ADDRESSES ABRAMSON’S ARTICLE. I HAD NOT SEEN IT AT THE TIME I WROTE THIS. it COVERS A COUPLE OF VERY IMPORTANT POINTS THAT I DID NOT COVER. ONE VERY INTERESTING THING IT SHOWS IS THAT AMONG ALL VOTERS NATIONALLY, CLINTON HAS WON MORE OF THOSE WHO DECIDED WHO TO VOTE FOR IN THE LAST WEEK. IT ALSO REMINDS US OF THE FACT THAT “WHEN YOU DECIDE TO VOTE” =/= “WHEN YOU DECIDED FOR WHOM YOU WERE VOTING”. IMAGINE THAT!
As you can see in the image to the right, it also lays out the case that late-deciders do not overwhelmingly break for Sanders in all cases. In 19 states where this was assessed via exit polling, he only did better with late-deciders in 11 of them. This isn’t enough to realistically claim there’s a correlation. In Georgia, Massachusetts, Missouri, Nevada, New Hampshire, South Carolina, Texas and Vermont, Sanders did worse with late deciders. Just to put it into perspective: had “Election Day” not occurred, Sanders would have won Missouri.
“Consider: in North Carolina, Hillary Clinton only won Election Day voting 52% to 48%. Given the shenanigans in evidence during the live voting there — thousands of college students were turned away from the polls due to insufficient identification under a new voter-suppression statute in the state — it wouldn’t be unfair to call that 4-point race more like a 2-point one (51% to 49% for Clinton).
OK, first of all: woulda, coulda, shoulda. You don’t get to revise the actual margins of a race and claim a different result because you believe that your candidate’s voters were disproportionately impacted by technical problems. The fact is that a federal court is currently evaluating whether North Carolina's voter ID law disproportionately impacts minorities. Guess which particular minority group comprises the overwhelming majority of minorities in North Carolina? Guess by what margins said minority group supports Hillary Clinton?
As the Washington Post outlined in a review of voter ID laws from the GAO, turnout in KS & TN among black voters dropped by 3.7 and 1.5 points more (respectively) than it did among white voters after voter ID was implemented; among 18 year-olds in Kansas and 19-23 year-olds in Tennessee, turnout dropped by 7.1 and 1.2 points more (respectively) than it did among those aged 44-53. Even though the percentage drop-off was larger among young voters, black voters comprise a larger share of the Democratic primary electorate than 18-23 year-olds in most states. When you break down what each particular group comprises as a share of the electorate, this means that a larger raw number of black voters are disenfranchised than college-aged voters.
Bottom line: it’s much more likely that more Clinton supporters were kept from voting in the NC Democratic primary than Sanders supporters due to voter ID laws.
Consider: on Super Tuesday 3, because early voting is always reported first, Clinton’s margins of victory were originally believed to be 25 points in Missouri, 30 points in Illinois, and 30 points in Ohio. Missouri, which doesn’t have conventional early voting, ended up a tie. Illinois ended up with a 1.8% margin for Clinton (after being a 42-point race in Clinton’s favor just a week earlier) and Ohio a 13.8% margin. Any one of us could do the math there. And yet the media never did.”
He sort of jumps around here. In one breath, it’s implied that he’s comparing early vote versus ED vote. In the next, he’s talking about polls (regarding Illinois). I’m assuming based on the margins he outlined for those three states that he is talking about the polls.
Michigan’s margins were off by that much according to most polls, yes. Illinois, Ohio and Missouri, however, were not. Polls conducted weeks or even months before the election are irrelevant; polls are a snapshot in time.
The final two polls of Missouri according to RealClearPolitics showed Sanders up by 1 and Clinton up by 7 (an average of Clinton +3; the result was Clinton +0.4; within margin of error).
For Illinois, the final three polls conducted in the 10 days preceding the election showed Clinton with an average lead of 2.3 points; she won by 1.8.
In Ohio, the average was Clinton +8; she won by 13.8 (still within the aggregate MoE for most polls). In this case, she actually exceeded polling expectations.
One of us did just do the math, and the claims made in the HuffPo piece are factually and demonstrably wrong.
Consider: in Arizona yesterday, the election was called almost immediately by the media, with Clinton appearing to “win” the state by a margin of 61.5% to 36.1%. Of course, this was all early voting. CNN even wrongly reported that these early votes constituted the live vote in 41% of all Arizona precincts — rather than merely mail-in votes constituting a percentage of the total projected vote in the state — which allowed most Americans to go to bed believing both that Clinton had won Arizona by more than 25 points and that that margin was the result of nearly half of Arizona’s precincts reporting their live-voting results. Neither was true. In fact, as of the time of that 61.5% to 36.1% “win,” not a single precinct in Arizona had reported its Election Day results. Indeed, more than two and a half hours after polls closed in Arizona, officials there had counted only 54,000 of the estimated 431,000 Election Day ballots.
First of all, the race was not called “almost immediately”. Arizona has a one-hour moratorium on releasing any results after the polls close. This ensured that all early vote was counted by the time the media began reporting vote totals. The better part of an hour after that, the race was called for Clinton. In reality, the polls had been closed for almost two hours before the call was made: hardly “immediately”.
The media made a series of mistakes on Election Night in Arizona, with the biggest one being the use of the words “precincts” and “percentage reporting”. Arizona used what we’ll deem “voting locations” for this primary, which did not correspond to precincts at all. Any voter could go to any voting location in his or her county to cast their ballot, regardless of where they lived. Results were not collected nor recorded by precinct. The media, however, used this term for whatever reason (presumably as a way to indicate percentage of the likely vote counted).
Furthermore, the statement that “only 54,000 of the estimated 431,000 Election Day ballots” had been counted more than two hours later is completely and utterly wrong. I have no idea where he is getting this number. I thought perhaps he was talking about both primaries. There were just over 940,000 preliminary votes cast in both primaries, with 70% of those being early votes (leaving approximately 280,000 votes cast on Election Day across both parties). In the Democratic primary, there were only around 410,000 total preliminary votes cast. This means approximately 125,000 Democratic votes were cast on Election Day.
Many people have a hard time with the meme that’s going around that says “Arizona was called with only ‘2% reporting’”. Again, the media was using terms like “precincts” and “% reporting” that night for Arizona that really did not apply like they usually do in other races; most states only cast 30 to 50 percent of their overall vote early and do so at their actual precincts.
Nevertheless, when the race was called nearly two hours after the poll closed, anywhere from 70 to 75 percent of the statewide vote had been counted (because of the huge dump of early voters). Based on the margin and the percentage of the overall vote, there was no way this was going to be surpassed by the outstanding Election Day voting.
Sidenote: While individual voter suppression is a deplorable thing, doing the math based on the figures to the right will tell you that unless close to 90% of the vote in line had been for Sanders, then there was no way the overall result was alerted by a call with outstanding lines. Proportional delegates do apply here; the early call couldn’t have made a difference for more than one delegate at most — if even that — due to the relatively small segment of the vote that was Election Day vote.
I’ll wrap this up by addressing the final chunk of text that is relevant to the overall theme here (the rest of it is regarding superdelegates and other, more abstract concepts):
So how did Bernie Sanders do on Election Day in Arizona?
As of the writing of this essay (2:45 AM ET), Sanders was leading Clinton in Election Day voting in Arizona 50.2% to 49.8%, with just under 75,000 votes (about 17.3% of all Election Day votes) counted.
So imagine, for a moment, that early votes were reported to the media last rather than first. Which, of course, they quite easily could be, given that they’re less — rather than more — reflective of the actual state of opinion on Election Day. Were early votes reported last rather than first, Arizona as of 2:45 AM ET would have been considered not only too close to call but a genuine nail-biter. In fact, only 400 or so Election Day votes were separating the two Democratic candidates at that point — though the momentum with each new vote counted was quite clearly in Sanders’ favor.
So the question becomes, why does any of this matter? Does the point being made here — that Bernie Sanders is as or more popular than Hillary in both all the states he won and many of the states he didn’t — gain Sanders a single delegate? Does it move him one inch closer to being President?
No.
What it does do is explain why the Clinton-Sanders race is a 5-point race nationally — just a hair from being a statistical tie, given the margin of error — despite the media treating Clinton’s nomination as a foregone conclusion.
What it does do is explain how Clinton is “beating” Sanders among American voters despite having a -13 favorability rating nationally, as compared to Sanders’ +11 rating. That dramatic difference is possible because in favorability polling, pollsters only count voters who say they know enough about a candidate to form an opinion. That eliminates the sort of “early voters” who cast ballots for Hillary Clinton before having much of a handle on who Bernie Sanders is.
Again, this is a case of “woulda, coulda, shoulda”. Why exactly should it matter in what order votes are counted if the overall result will become apparent in a matter of hours? He basically asks this himself, but doesn’t provide a substantial answer. There is a lot of lamenting about how people can vote early — rather than waiting until the last minute — with the suggestion that hordes of people would change their minds if they only had a couple of weeks more to decide. Furthermore, there is no credible polling or election results that suggest this is a “5-point race nationally”. It is currently a 18-point race in actual votes and a 9-point race in aggregate polling.
The real answer to his perceived problem is one he clearly mentioned at the beginning of the piece: younger voters disproportionately vote late, younger voters overwhelmingly support Sanders, and this is why the Election Day vote is more pro-Sanders. It isn’t a case of people digesting more information and making different decisions en masse. He clearly acknowledged that younger voters cast their ballots on Election Day more so than other groups in the first couple of paragraphs, but doesn’t make the connection because it doesn’t suit the overall narrative.
Even though his premise is wrong, it still manages to be condescending in a way that has been present in many pro-Sanders circles and arguments: that non-Sanders supporters are ignorant and need education (or even worse, are conservative hacks). Whether it’s applied to older voters, female voters, black voters, or Clinton supporters as a whole, the implication is always one of “Well, if those people were only more educated and enlightened, then they’d be voting like us”. In this case, Mr. Abramson’s article suggests that voters need more time to be properly educated by seeing television commercials and by being within spitting distance of a candidate.
While he doesn’t come right out and say it, the article carries a whiff of there being a desire for voters to not be able to vote early. This is the same thing that the GOP says and feels about general elections: that the uneducated hordes of Democrats stuff the ballot box at their convenience ahead of the election, and that they’d prefer “more educated voters” to vote instead. Why should it matter if a candidate is winning Election Day vote if they are losing the overall vote? A vote is a vote is a vote.
In the words of Mr. Abramson:
Why does any of this matter?
It doesn’t, unless you’re merely being a sore loser. Extracting “wins” from objective losses, implying your opponents’ supporters are ignorant, implying that the time-line in which they can cast their ballots should be restricted and using faulty numbers are all big no-nos in my book. They’re also what the GOP does: not us.