To do this, I've gone through six months of crosstabs from biweekly YouGov / Economist polling, focusing on black voters in particular (since they're likely to prove the decisive constituency).
First, let's look at three crucial indicators for Hillary Clinton among black voters: name recognition, preference (versus other candidates), and favorability:
Most of this should be fairly straightforward and uncontroversial. Hillary's high name-recognition is a testament to several decades of national prominence, largely beginning with the presidency of Bill Clinton in the nineties. This is more of a test of political literacy than anything: the 9% who answered "don't know" when asked about her matches the minimal 9% who will generally answer "don't know" about any given political topic.
During this period, black voters have developed a fairly consistent opinion of Clinton: about 70% of them view her favorably, though that number has occasionally slipped a few points over the past half year. Crucially, however, the Clinton indicator that has shown the most variability among black voters has been actual preference. Though this number is back to where it started at around 64%, it has dropped to as low as 53% and climbed to as high as 77%.
The picture this paints of Clinton is fairly intuitive, though not quite as flattering as her advocates often suggest. Black voters overwhelmingly know her, and a significant majority of them like her - two factors that don't seem particularly variable over time. Nevertheless, a minimum 23% and as many as 47% of black voters have on occasion refused to back Clinton for president. Those are the numbers we should begin with when we consider the support of black voters for Bernie Sanders.
When we look at the same three indicators for Sanders, another straightforward trend appears:
Across the board, Sanders' numbers have steadily increased among black voters throughout his campaign, typically climbing around 1% every week. You could probably come up with all sorts of elaborate circumstantial explanations for this involving various improvements in Sanders' platform, but this does little to explain the sheer consistency of his ascent. The simpler and more compelling explanation here is that Sanders started with relatively low name recognition among black voters, and that the more black voters get to know him, the more they like him.
There is of course no way to know if and to what extent this upward progression in Sanders would continue, given enough time. If we compare his numbers to Clinton's, however, I think we can begin to draw some basic conclusions about how black voters have aligned themselves in the Democratic primaries.
Competing for black voters
Here's how two of these trends compare for each candidate:
This chart seems to affirm quite clearly the theory just given: Sanders has primarily suffered from a lack of name recognition among black voters. His numbers are rapidly aligning themselves with Clinton's; at the current pace, his favorables would be identical with Clinton's by April, with name recognition for both candidates maxing out at around 90%. If we look at cumulative candidate preferences, however, a few separate trends emerge:
When we note that Biden dropped out of the race on October 21, a fairly clear story emerges about the Democratic primary.
Clinton's name recognition among black voters is essentially maxed out. This has translated into a favorability rating of 70% and a base of 55% support among black voters. Both show signs of steady but minor deterioration; however, in late October, Clinton received a massive windfall of orphaned Biden supporters. This faction of her coalition - accounting for about 20% of the black vote - originally hoped to vote against her; it's not clear how strong their support really is.
Sanders, meanwhile, began the race with an extraordinary name recognition disadvantage among black voters, which appears to correlate directly with deficits in favorability and support. As his name recognition has grown, both numbers have grown proportionally.
If these trends over the past six months continue, it appears likely that Clinton's numbers would continue to deteriorate to her pre-Biden surge levels in the mid fifties, while Sanders' numbers would continue to climb into the mid-thirties. Crucially, while this still marks a significant preference among black voters for Clinton, it would not be enough to provide the so-called "firewall" she's relying on to win the nomination. For instance, all else being equal in Nevada, a 54-35 split of black voters would have been enough to hand Sanders the victory.
Without making any further predictions, it seems clear that name recognition has been Sanders' greatest disadvantage among black voters. His improvement in the polls seems almost entirely a matter of improved familiarity among black voters, which suggests that he may have started as many as 30 points higher if he were better known. That deficit also left him poorly positioned to pick up orphaned black Biden voters, who broke 2:1 for Clinton and gave her an additional 10 point advantage.
In my next post, I will discuss how the role of black voters in the primaries has been largely misunderstood by the media - often, of course, in the most reactionary way possible.