Here is the thing. The data has always suggested the Hillary was vulnerable in Michigan. I know that is hard to believe since the polls said that she would win by 22 points. But the polls were making amateur mistakes. And it was preventable. Turnout, in states outside of the southeast, has been steadily rising. People dismissed it because they were caucuses. Even I worried how primaries would go. But if you followed the details you saw that Bernie is generating turnout and had generated more and more turnout with every win. You can see my earlier diaries on the topic here and here.
Anyhow, the signs were there. Just like many on this site, there was an assumption that if Hillary was winning the African-American vote by 50, 60 points in the southeast, she would win it by similar margins elsewhere. But the margins of African-American voters in the southeast weren’t even identical. Markos himself made this mistake the other day:
Mississippi is by far the most African-American state (by percentage) in the union. Runner-up Louisiana is 32.4 percent black, while Mississippi is 37.3 percent black. So, I would guess that Clinton’s victory will be even wider than her 71-23 percent blowout in the Bayou State.
Look, his prediction was right, but not for the right reasons. There was no direct correlation between the % of black voters in the southeastern states and the margin of victory. It certainly predicted that Hillary would win, but everyone knew that. Here is the chart I shared yesterday that helps show why.
State |
Winning Margin |
% Black |
Louisiana |
47.9 |
33% |
Georgia |
43.1 |
33% |
South Carolina |
47.5 |
29% |
Alabama |
58.6 |
27% |
As you can see, they all went by big margins, but while Georgia was tied for the highest percentage of black voters, the winning margin there was 15 points lower than Alabama — which had the highest margin of victory and the smallest black population. The African-American vote, like the votes of other demographics, are not the exact same everywhere.
In simpler terms though, there is something clearly off when national polls are tightening and nearly all of the state polls are expanding. And broken turnout models helps explain that. And that’s the purpose of this diary. To continue to show how turnout is predictive of results in this year’s primary — to some limited degree irrespective of demographics. And with Michigan and Mississippi now in my poll, you can see that both continue that trend.
State |
2008 |
2016 |
Change in Turnout |
Winner |
Winning Margin |
Michigan |
594,398 |
1,183,840 |
99.17% |
Sanders |
1.7 |
Kansas |
37,089 |
39,043 |
5.27% |
Sanders |
35.4 |
Maine |
44,670 |
46,800 |
4.77% |
Sanders |
28.8 |
Colorado |
120,001 |
124,000 |
3.33% |
Sanders |
18.7 |
Minnesota |
214,066 |
215,000 |
0.44% |
Sanders |
23.2 |
Massachussets |
1,263,764 |
1,209,000 |
-4.33% |
Clinton |
1.4 |
New Hampshire |
288,672 |
254,776 |
-11.74% |
Sanders |
22.4 |
Vermont |
155,279 |
134,571 |
-13.34% |
Sanders |
72.5 |
Nebraska |
38,670 |
33,460 |
-13.47% |
Sanders |
14.2 |
Oklahoma |
417,207 |
335,554 |
-19.57% |
Sanders |
10.4 |
Virginia |
986,203 |
784,392 |
-20.46% |
Clinton |
29.1 |
Louisiana |
384,346 |
287,000 |
-25.33% |
Clinton |
47.9 |
Alabama |
536,626 |
397,890 |
-25.85% |
Clinton |
58.6 |
Iowa |
236,000 |
171,109 |
-27.50% |
Clinton |
0.3 |
Nevada |
117,559 |
84,000 |
-28.55% |
Clinton |
5.3 |
Georgia |
1,060,851 |
755,000 |
-28.83% |
Clinton |
43.1 |
Arkansas |
315,839 |
224,000 |
-29.08% |
Clinton |
36.6 |
South Carolina |
532,151 |
369,526 |
-30.56% |
Clinton |
47.5 |
Tennessee |
624,764 |
371,321 |
-40.57% |
Clinton |
33.7 |
Mississippi |
434,110 |
220,159 |
-49.28% |
Clinton |
66.1 |
Texas |
2,874,986 |
1,430,000 |
-50.26% |
Clinton |
32.0 |
Note right up front that Michigan’s turnout change is through the roof. If you don’t recall, the 2008 primary in Michigan was a mess. Every candidate had withdrawn except for Hillary who beat “uncommitted” 56 to 40%. It was bizarro land. It somehow was still one of the best turnouts in Michigan Democratic primary history (from my informal searches) but I suspect 2008 turnout would have been significantly larger in 2008 without the weirdness. So the 100% increase in turnout significantly overstates the turnout bump which is why it looks like an outlier there. No doubt it was a big turnout bump, but it’s impossible to say how big (comparatively speaking anyhow).
Now here is how the delegates doled out by turnout break down.
Delegates To Date
|
Hillary |
Bernie |
766 |
549 |
Greater than 2008 Turnout |
139 |
192 |
0-20% Less than 2008 |
119 |
126 |
More than 20% Less than 2008 |
508 |
231 |
- The 0-20% drop in turnout states include Massachusetts of course.
- I’ve excluded American Samoa for now. I’ll work non-states into the next update on delegates.
There won’t be many more uncontested states i’d think. So don’t expect there to be many more Mississippi’s. I’m not sure what Bernie is investing in North Carolina though so that might be the biggest margin remaining on the board for Hillary. I think Missouri should be contestable, but it’ll depend on investment. Here is a simplified break down of where delegates are coming up:
States |
States Completed |
States to Go |
Blue States |
339 |
1661 |
Swing States |
264 |
357 |
Red States |
712 |
599 |
Non-States |
6 |
113 |
|
1,321 |
2,730 |
Note that we are about 1/3 of the way through the contest. Blue states so far have been either competitive or big wins for Bernie. Also, the red states that are left to go primarily are the ones where Bernie is projected to do well. But the toughest road is going to be the next week. The question will be is how big of a win is North Carolina for Hillary? And does Bernie close the gap in Florida and does he win Ohio?
One last point. When comparing delegates to 2008, which I don’t do here, make sure to weight your delegates appropriately. In 2008, there were 3,253 pledged delegates and this year there are 4,051 pledged delegates. So there are a lot more delegates to win at the polls. This does not include super-delegates which are meaningless unless it’s super close.
All delegate data herein is based on 538’s tallies. I may switch from that as they try to get ahead and calculate delegate proportions before official proportions are released and they’ve been wrong on multiple occasions. Still, going with it for now. projects.fivethirtyeight.com/...