With the completion of the Wisconsin primary, in which Hillary Clinton yet again did just a little better than she did in 2008, it’s time to update the Democratic primary comparison graph. We first took a look at how this year’s Democratic primary is shaping up compared to 2008 back in March. Conveniently, from an analysis perspective, we have the same candidate running in both years. We now have a lot more contests to put on our graph. Above, we see Clinton’s vote share in 2016 compared to 2008.
Where Clinton is doing better this year, the circles are above the diagonal line. The area of each circle is proportional to the number of 2016 delegates. For a true apples-to-apples comparison, certain states have been excluded, such as states where John Edwards won more than 10 percent in 2008.
It’s clear that most of the time, Clinton has a higher vote share in 2016 compared to 2008. Considering that with the full inclusion of Florida and Michigan this year, she only needed to do a little bit better than 2008, that would be enough to win, if she maintains this pattern through the end of the primary season. However, her massive improvements in Southern states gave her such a large delegate buffer that she could even afford to do worse than 2008 in many states and still win.
Even states where we don’t have a good comparison contest in 2008 (Florida, for instance) count towards the delegate total, though—so you can still see them here, and below the fold.
For a closer look, we’ll start in the South, which is mostly complete. In all but two states, Clinton improved by leaps and bounds over her performance in 2008. Those two states are Arkansas and Oklahoma. As discussed in depth previously, large proportions of the Democratic electorate in these two states voted for a candidate other than Obama in the 2012 Democratic primary, when Obama had no serious challenger. Going forward, we expect Kentucky and West Virginia to follow this pattern as well.
For the rest of the Southern states, Clinton’s increased performance can be attributed mostly—but not entirely—to varying levels of increased support from African American voters. It’s important to note that black voters were not unanimously supportive of Obama in 2008, and they are not unanimously supportive of Clinton this year, although it does come close in a few states, most notably among older voters.
Below, we can see how Clinton’s share of the vote has changed as the share of the Democratic electorate which is black increases (this includes non-Southern states). While there’s clearly a relationship, there’s also a lot of room for additional factors to come into play. Compare, for example, Arkansas (home of GOP-voting ancestral Democrats) to Virginia (home of a good chunk of the DC beltway). While in both states, there was a massive swing of black voters to Clinton’s side, in Arkansas Clinton lost some support from white voters, while in Virginia Clinton gained ten points of support. (This is something to keep an eye on in Maryland’s upcoming primaries.)
Setting aside Illinois (Barrack Obama’s homestate) and Iowa (where Edwards won a large share of support in 2008), Clinton has shown pretty consistent improvement over 2008. Her improvement was a little greater in the earlier states (MN, KS, NE). This could be related to their format (caucuses), specific demographics, or a small amount of momentum for Bernie Sanders in later voting states.
Overall, it is very surprising how close Clinton’s numbers are in both 2008 and 2016 in some states, considering the changes in support among different demographics as discussed above. Consider, for instance, the results maps of Ohio:
Simply put, there’s massive changes in support by geographic region. Yet in the end, Clinton’s support statewide was very similar to 2008, just three percentage points more. Falling support in rural areas was coincidentally offset by rising support in the most populous counties.
We can also see dueling trends among demographics in the exit polls. Below, we have the support for Clinton by age from the recent Wisconsin primary compared to 2008. Clinton’s support rose among older voters, but fell among younger voters. Overall, the two processes mostly cancelled each other out, and Clinton’s statewide vote share increased by just 2.3 percentage points.
As additional examples, you can see the exit poll comparisons by education for Wisconsin, and by both age and education for Ohio, as well.
Repeating the exercise for Western states, Utah immediately jumps out. Clinton’s support fell drastically from 2008 to 2016, by nearly 19 percentage points. What happened in Utah?
Simple: they switched to a caucus.
We also see that Clinton fared worse in both the Washington and Alaska caucuses compared to 2008. We might have expected her to do worse in Idaho, too, then, but this is not the case. At any rate, it would not be a surprise for Sanders to outperform Obama in Oregon’s May primary.
There have been no contests in the Northeast since the last post, but I’ll add the graph in here anyway for the sake of completeness. As New Hampshire was a three-way race in 2008, and Vermont is the home state of Sanders, that only leaves us Massachusetts and Maine that are apples-to-apples comparisons. In both states, Clinton did a little worse in 2016 than in 2008.