Cos (cos) wrote,

What about Michigan?

Since my post concluding that Clinton will get the Democratic nomination, people keep asking me if I think anything changed after Bernie Sanders' surprise upset win in Michigan.

First of all, I'm overjoyed that he won! I'd love to see him continue to do well in future primaries, and I think it will be a very good thing for the Democratic party and for the country if he does.

Second, this makes me feel like I should not have looked at the polls, and just written based on what I said was doing: Looking at the states that already voted, and extrapolating the same voting patterns. Polls shortly before an election are usually pretty good at predicting the results, but actual voting in primaries has long seemed to be a much more solid predictor of subsequent primaries. And based on that, Michigan's result was not at all surprising. Industrial midwestern states' Democratic primaries are similar to Massachusetts but a bit better for Bernie, while mid-Atlantic and southern New England states are a little worse for him. Based on Super Tuesday, I would've predicted he'd probably lose the mid-Atlantic and southern New England states, but win most of the industrial midwest by small margins (with a bigger margin in Wisconsin, and PA being the one he's most likely to lose).

However... I actually did most of my analysis for the last post before looking at the Michigan polls. Sure, I knew that Clinton led him by a lot in the midwest in polls earlier in the season, but I assumed that lead would disappear like it had in plenty of other states where she started with a big lead. Early polls very often favor the bigger name in the race. In other words, that post reflected my conclusions based on a belief that Sanders would probably win Michigan narrowly. Then I looked at the polls, and to my surprise Clinton's lead was still very large on three polls released just in the past week (as of the day before the Michigan primary). So I changed my wording a little bit to reflect that.

For me, it wasn't so much Sanders' narrow win in Michigan that was a surprise, it was the polls! Michigan's actual result was very much what I expected until the last couple of days, when I looked at those polls before finishing my post.

That said, let's look at the numbers a bit more specifically...

Since Super Tuesday:
Kansas: Sanders 23, Clinton 10
Nebraska: Sanders 15, Clinton 10
Maine: Sanders 16, Clinton 9
Louisiana: Sanders 14, Clinton 37
Michigan: Sanders 67, Clinton 63
Mississippi: Sanders 4, Clinton 32
Total: Sanders 139, Clinton 161

On the same day Sanders' remarkable upset win in Michigan netted him 4 more delegates than Clinton, her long-expected small-news overwhelming win in Mississippi netter her 28 more delegates than him. In total that day, 166 more delegates were determined. Since Super Tuesday, 300 pledged delegates were determined, leaving only 2724 to go. Clinton's lead is now 221.

(Note: I'm using numbers from, who I've found have better calculations of how primary and caucus results will translate into state delegates than much of the press. But the differences are slight, so it doesn't make a big difference.)

What the above numbers highlight is what really stood out when I looked at Super Tuesday: It's not just that Clinton was winning all the southern states, it's the scale of her wins. Sanders' huge 22-point win in New Hampshire made big news, and he went on to score very solid wins in caucus states - 19 points in Colorado, 22 points in Minnesota, 14 points in Nebraska 29 points in Maine. While this was happening, Clinton scored astounding blowout wins in the much more populated southern states: 59 points in Alabama, 66 points in Mississippi, 48 points in Louisiana, 43 points in Georgia, 47 points in South Carolina, and a mere 36 points in Arkansas. Clinton literally beat him more than 4-to-1 in several states, and even 5-to-1 in a couple. The only place Sanders got a comparable win was his home state of Vermont.

These wins didn't make as big a splash in the news as New Hampshire and Michigan, because Clinton was expected to win these states all along. The news is much too focused on who "wins" each state, and it's misleading. Democratic primaries award delegates proportionally. The difference between winning 51-49 vs. losing 49-51 is a huge deal for the press, but it may be nothing in terms of delegates (as in Massachusetts, where Clinton "won" but Sanders and Clinton will get 45 and 46 delegates each; often a 51-49 win results in an even split). The difference between winning 55-45 vs. 59-41 may actually be bigger in terms of delegates, but won't have much effect on how the press talks about it.

If I'd looked at the numbers after Super Tuesday and seen Clinton ahead by about 100 delegates, I would've written a very different post. I could clearly see a path for Sanders to make up a 100 delegate difference, and that path did indeed lie through winning the midwestern states like Michigan, not necessarily by a lot. Let's look at that path:

An optimistic scenario for Sanders that's consistent with the voting patterns we've already, has him continuing his winning streak in caucus states, with results ranging from roughly high 50s to low 70s. Let's say he averages 65% of the delegates in remaining caucus states: out of 279 delegates remaining, he'd get about 181 and Clinton would get about 98, for a gain of 83.

Sanders is likely to lose southern New England and the mid-Atlantic states mostly, but not by a lot, except in New York where Clinton's home state advantage could give her a 60-40 win. Let's say he holds her to to reasonably small victory margins overall, and wins 46% of the 468 delegates (CT+RI+NY+NJ+DE+MD), 261 to 307 for Clinton, a loss of 46.

He can mostly make that up with modest wins in the remaining industrial midwest. Pennsylvania is part industrial midwest / part mid-Atlantic, and Indiana is similarly half-southern, which limits his ability to win these states, but lets say he manages to narrowly win PA and lose Indiana by only a little, and has more solid wins in Ohio and Illinois. Those states total to 571 delegates, so he'd likely gain about the same as Clinton does in southern New England + mid-Atlantic.

In the upper midwest and great plains past Appalachia, I expect Sanders to win most states comfortably. Several of these, plus much of the mountain west, are caucus states which I already considered a few paragraphs at all, but the primaries yet to come in these regions are Missouri, Wisconsin, Montana, South Dakota - 202 delegates. Since we're doing the optimistic Sanders scenario, let's give him 60-40 over this set as a whole: 121 Sanders, 81 Clinton.

But there are still some southern states left to go! North Carolina, Florida, West Virginia, and Kentucky. West Virginia is a bit odd because it's so white compared to the rest of the south, but on the other hand Appalachia was by far Clinton's best region in 2008 and it has been very good for her this time around too. I honestly can't predict how WV will turn out, but it's only 29 delegates. Florida and Kentucky may be more modest wins for Clinton than other southern states... but that's not saying a lot. Let's split these 405 delegates 60/40 for Clinton - I think that's generous to Sanders, but plausible. So 243 to 162, for a Clinton gain of 81.

Arizona and New Mexico... Sanders didn't even win the one southwestern caucus so far, Nevada. But Clinton's campaign had been organizing deeply there for a much longer time, which matters in caucuses, and most of that organization happened before New Hampshire. I wouldn't be surprised if she wins these primaries too, but Sanders might (well, New Mexico isn't until June 7th, so the campaigns may have ended by then, but if Sanders were going to win in this scenario, he'd need to keep going until then). Of these 109 delegates, let's say optimistically that Sanders 59 to Clinton's 50.

The Pacific Northwest should be Sanders' best region outside of northern New England. With a strong campaign, he can win big there. Washington is a caucus, already included in the numbers above; if Sanders wins Oregon by 70%+, those 61 delegates could split something like 44-17.

Caucuses:	S 181	C  98	Sanders +83
NE/Mid-Atl:	S 261	C 307	Sanders -46
Mid/Plains/W:	S 121	C  81	Sanders +40
Remaining S:	S 162	C 243	Sanders -81
Southwest:	S  59	C  50	Sanders  +9
Oregon:		S  44	C  17	Sanders +27

Total: Sanders +32.

He could do better even better than this. Maybe he wins Pennsylvania and even Indiana, maybe his caucus results average 70%, maybe he comes close to a tie in the non-NY mid-Atlantic states. I think I've described a fairly optimistic scenario for Sanders here, but we could ratchet it up some more and end up with Sanders +50 or even +60.

Which leaves us with California, 475 delegates! I'm having a very hard time extrapolating current voting patterns to California. West coast is likely to be Bernie-friendly. Clinton has a strong advantage in states with a long-standing dominant Democratic party establishment (like southern New England). California has a large latino population which has trended more towards Clinton than Sanders. But it has fewer conservative Democrats than the east coast Democratic states, which are institutionally Democratic; in California, fewer conservative people are Democrats. I can imagine reasons to predict the state goes either way. I doubt it will be a *huge* victory for either candidate, but even a modest victory in a state with so many delegates can be significant.

If Sanders were at a 100 delegate deficit, and he performed really well after Super Tuesday, he might cut that deficit down to as little as 40 delegates (but more realistically 60-something). Could he make up the difference by winning 475-delegate California? Sure, maybe. California votes June 7th, and if Sanders were that close to Clinton that far out, which would be astonishing, people might feel the momentum is with him, and California might really turn out for him.

But Clinton is leading by 220, not 100. She didn't just win the south, she won a series of overwhelming blowouts, and her lead in delegates already elected is just too large. Barring some huge shift in the dynamic of the campaign, Sanders isn't going to be able to cut that lead down to even 100 before California, let alone 60. He'd need to win by almost 80% in California to gain over 100 delegates, and chances are he'll be well over 100 behind.

Will Sanders win more states than Clinton for the rest of the campaign? Perhaps he will. Is he going to gain more delegates than Clinton from now on? He very well may - and in fact, I think it's quite likely. I expect him to cut down on her lead, and I very much want him to.

Sanders has already surprised the political world by showing he's a real and credible candidate who had a chance to win. With each win from now on, and with each gain in delegates cutting into Clinton's lead, he'll continue giving people the perception of an even race. Clinton will have to keep trying, really trying, in order to keep earning the delegates she needs. She cannot relax, and the campaign remains real. At the end, Sanders will have accomplished a lot, and helped bring about a major shift in the Democratic party, one that we can keep building on.

But Clinton doesn't need a big lead to get the nomination. If she goes to the convention with even one more pledged delegate than Sanders, she will be the Democratic nominee. Will Sanders cut Clinton's lead down all the way from over 200 now, to 0 at the end? Unless some big unexpected surprise (bigger than beating the polls in Michigan) shakes up the race, then no, he will not.

Edit: Republican primary
Today (March 15th), we find out whether Trump wins both Florida and Ohio, giving him a good shot at winning a majority of delegates and the nomination, or whether he loses one of them (or maybe both), making it much more likely that he won't get a straight majority, and the Republican convention will be contested.

  • We are hiring a lot more tech people, want to work here?

    The department I work for is looking for software developers, cloud engineers, SREs, product managers, and probably a bunch of other roles I haven't…

  • Paid in full

    I just got today, in the mail, my "paid in full" papers for a mortgage I refinanced (and thus paid off in full) in August 2019. Crossposted from…

  • Job opening in my group

    The group I'm in has a job opening for Lead Site Reliability Engineer (SRE). This is to replace someone I worked with who left to join a company…

  • Post a new comment


    default userpic

    Your reply will be screened

    Your IP address will be recorded 

    When you submit the form an invisible reCAPTCHA check will be performed.
    You must follow the Privacy Policy and Google Terms of use.