WHAT TO EXPECT: Super Tuesday 2: Electric Boogaloo -- Clinton still favored

I am returning to write a bit about the primaries on Tuesday, starting with a brief guide to Illinois, which is, along with Ohio, the big prize tomorrow that is up for grabs, though Hillary seems favored in both (a cautious estimate). I’m a little weary by the intensity of the debates and the instant outrage machine that floods my facebook and this site after every relatively minor mistake or unforced error for Hillary Clinton. I think the lovely Larry David had it best in the other days SNL skit, “I mean, I’m great. But I’m not five-posts-a-day great.”  So I’m not going to use this post to rehash any specific political arguments, and I ask that neither Clinton supporters nor Sanders supporters do so in the comments. Instead, please share observations if you live in one of these states, your own predictions, and discuss the actual elections coming up, and leave the politics for the other thirty something threads about it going on at any time.

So starting with Illinois, I quickly pieced together this little table (using the 2010 Census figures), based on the main demographics and variables that have been at play, at times, in the 2016 primary thus far.

Illinois (156 delegates):

This chart should make obvious just how Cook County (Chicago and some immediate suburbs) dominates the state, and even more-so dominates the Democratic primary. Nearly 50% of all votes cast in 2008 came from Cook county and this is fairly consistent. Basically what it means is, if you look at the results and Clinton has lost every county in Illinois 55-45, and then wins Cook 60-40, she wins Illinois. Illinois has open primaries as well, and I expect that given the hard-right trend in Madison county and many of the rural southern IL counties, that those areas will cast considerably fewer votes this year, with many flocking into the Republican primary, meaning Cook will probably have even more influence this year.

Polling is pretty noisy around Illinois. Not least because there was only scarce polling, and now we’ve had a couple of polls all come out in one week, some of them showing radically different results. First, I will toss out We Ask America, because they are a garbage right-wing pollster and I’ve learned not to trust them (their poll had Hillary up 62-25 btw, so I’m not being partisan against Bernie). Chicago Tribune/Research America, also last week, had Hillary up 67-25 through the beginning of March (and I’m not as familiar with this pollster and Nate Silver doesn’t have it evaluated, so I will just leave it as is). Then this week we’ve had Marist, with Hillary up 51-45, and YouGov with Sanders up 48-46.

The main advantage going in for Sanders is that unlike New York, Pennsylvania, and Florida, Illinois is, like Michigan, an open primary. That means anyone can vote. In Michigan Hillary won the vote of self-described Democrats by double digits, but there was strong turnout in the primary of independents and third-party registrants and they gave Sanders 70% of the vote and single-handedly led him to victory in the Michigan primary, a dynamic that could return in Illinois, particularly if we again see a number of Democrats voting in the Republican primary to stop Trump.

My general feeling, looking at a lot of different factors, that Clinton is favored in Illinois. I think it will be a lot narrower than the Chicago Tribune pegged it, and between the Marist and YouGov polls, I trust Marist more, even though they were among those who missed in Michigan they have a good track record over all in the Democratic primary. Their 6 point margin is about where I was putting the race. The reasons why are of a more general sort, mainly that Illinois has a much stronger and better organized Democratic party with a more powerful Democratic establishment. The backing of party leaders and unions will go a lot further in Illinois than in Michigan, with it’s maligned and much weakened Democratic party. Particularly in the heavily urban Cook county, the local party organizations, led by leaders like Speaker of the House/IL Dem Chairman Mike Madigan, definitely seems to be working on the ground for Hillary Clinton, who in Cook county also has both a very big wedge issue in gun control, but also the backing of some very respected community leaders who can and will turnout voters for her, including Cook County Board President Toni Preckwinkle (a progressive black woman who I think will be the state’s next governor), Alderman Danny Solis, Congressman Bobby Rush, and Congressman Luis Gutierrez, who is a national leader in the immigration reform movement, and hugely influential in the Chicago Latino community. Combined with the extensive network of Unions doing GOTV for the Clinton campaign, and Obama’s almost-endorsement of Hillary Clinton, and Cook County should help her win Illinois.

Complicating matters for Sanders, is that Hillary won Oakland county in Michigan by 5 points even as she lost statewide by 1.5.  This is important because Lake, DuPage,  Kane and Will look most like Oakland county, only slightly less white and with a higher median household income. So to win Illinois, Sanders needs to do significantly better in big, wealthier, suburban counties than he’s done anywhere else. I’m of the opinion that the primary has fallen into a pattern and that the polling may move around a lot, but basic fundamentals end up mattering, just like in 2008 where the states typically followed fundamentals of demographics. In I have no doubt that Sanders will dominate everything outside of Chicago and its suburbs, save St. Clair county where black voters are around 50% of the Democratic primary electorate. Rock Island, Sangamon, Peoria, Champaign, Vermillion, Madison, most all of the rural counties (save a few of the southern flavored counties in the deep south part of the state that might still pull the lever for the Clinton brand). He’ll probably win McHenry county too. It’s just that wins in these areas aren’t going to be enough to outweigh Cook county, particularly not without a truly high-water mark of support in all of the suburban ring counties. And additionally, unlike Michigan, Hillary’s campaign doesn’t seem to be taking Illinois for granted. Not only is she spending more money, but both she and Bill are hitting the campaign trail heavily in the state in the lead up, and Priorities USA is also airing ads and spending millions on GOTV efforts. The campaign lost Michigan not the least because of overconfidence and because it let Sanders out-hustle them in the state in the critical final few days of the campaign, and seems determined not to make the same mistake in delegate rich Illinois.

The biggest problem for Sanders that I see in Illinois is that there isn’t a vote rich region like Western Michigan (Kalamazoo-Ottawa,Muskegeon-Kent and most of the surrounding satellite counties) that can weigh in as many votes for him. There are a few secondary urban centers he will dominate and probably gin great turnout in, and a lot of rural areas where plenty of angry conservative Democrats and anti-establishment voters will vote for him in protest against Obama, like they’ve been doing almost everywhere, but those areas will still combine to amount to a relatively small percentage of the primary vote, which is again, dominated by Chicago and a series of wealthy, yet still fairly diverse, suburban counties (the counties in the table above accounted for nearly 70% of votes in the 2008 Democratic primary).

Ohio (143 delegates):

More polls here, and the movement has been consistent and well-documented. A lot more polls here than either Michigan or Illinois, which makes me more confident. We’ve had five polls since March started. Here they in order from earliest to most recent:

CNN— 3/2 - 3/6 —294 LV — 63-33 Clinton

PPP (D)— 3/4 - 3/6 — 508 LV —56-35 Clinton

Quinnipiac— 3/2 - 3/7 — 521 LV —52-43 Clinton

NBC/WSJ/Marist— 3/4 - 3/10 — 453 LV —  58-38 Clinton

CBS/YouGov— 3/9 - 3/11 — 750 LV — 52-43 Clinton

Now, with Ohio, I can say that Quinnipiac has, in my experience, had a very good track record (unlike Quinnipiac in Colorado, where they’ve had a very bad track record since they started polling there). Nate Silver’s pollster scores have CNN/Opinion Research at A-, YouGov at C+, Quinnipiac at B+, and Marist at B+. Michigan had a lot of utterly shit pollsters polling it, including Mitchell (which gave Clinton huge leads) which is a crappy pollster that I never tend to trust, even though its accidentally right sometimes, even with its continually huge over 60 year old samples, and ARG! (the pirate pollsters). EPIC polled a little too early and missed any late movement to Sanders in the last week, Marist also polled a little too early, and had a flat out miss, (something which, statistically, will happen in a certain number of polls, because even the best pollster will sometimes produce results that are an outlier). Looking at the other “good pollsters” YouGov and Monmouth both were only about 7 points off the final results, or only slightly outside their MoEs.

Because Michigan was a surprising result, I’m more cautious on Ohio. I think Ohio is going to be very close. Either side could win it, but in a toss-up demographically, I’m going to go with the only other concrete evidence I have and side with the person the polls have up, but I think Hillary will win by between 1 and 4 percentage points. While polls can be suspect, and can be off, I think its wrong to automatically assume Sanders will always outperform his polling average by a large margin; Republicans were wrong back when they were “unfixing” polls in 2012 showing Obama ahead in all the swing states, and I caution against making the same mistake here. The semi-closed primaries in Ohio may save her, by keeping Republicans out of the Democratic primary and Democrats out of the Republican primary. I keep returning to fundamentals, but the 2008 primary serves as a pretty good guide. In the absence of demographics (New England, high % white liberal, bad economy), the 2008 primary has pretty accurately shown where each candidate plays best, though in a lot of areas, Hillary is winning Obama’s 2008 counties and losing hers.

Michigan comes in her as a outlier because we don’t know how it would have really voted in 2008. Hillary won it with 54.6% of the vote, but this was as she ran against “uncommitted” and was, along with Kucinich, the only name on the ballot. Neither Obama nor Edwards were contesting the state, which had had its delegates revoked at the time because of scheduling shenanigans. So Hillary’s almost 55% in a low turnout uncontested state where she was the only big name on the ballot doesn’t tell us how the state would have voted in a one on one between Obama and Hillary with both contesting the state. I have a suspicion that Obama would have won. In any case, with Ohio in 2008, the primary had a huge 2.2 million votes cast in a very high turnout primary that Hillary won by almost 10 points. Obama won just 5 counties in the state, albeit 3 of the most populous counties by comfortable margins.

This time, I’m expecting the map to look fairly different. Sanders will pick up Athens county (university of Ohio) by a huge margin. I expect him to flip Lucas county (Toledo, and not by a big margin, but probably by around 6-8 point margins) and Lorain by a similar margin. He’ll expand on Obama’s margins in Montgomery (Dayton) and in the Columbus suburbs and pick a wide swath of rural counties from Clinton in western and eastern Ohio. On the other hand, Clinton will flip Cuyahoga (Cleveland) and probably win it by Wayne county-ish margins, flip Hamilton narrowly (Cincinnati) and flip Franklin (Columbus). She should hold up pretty well in surrounding areas of northeast Ohio, like Summit, and the Mahoning area where she won 2:1 last time around, as well as the far southern rural counties like Scioto where she took 81% in 2008. Ohio will almost certainly be the closest state contested on Tuesday night and the most exciting state to watch. It’s also the state Sanders needs to win most to maintain his Michigan momentum.

Missouri (71 delegates):

This is a much smaller state. The only poll we have, exact only, had Hillary up 47-40, but I think Sanders is favored here. I imagine Hillary will do very well in Jackson and St. Louis City, and narrowly win St. Louis county, do well in the Mississippi river counties in the southeast corner of the state, and get manhandled everywhere else, including Springfield, Columbia county, where the University of Missouri is, and Jefferson county, the economically depressed, blue collar county that was once a Dem stronghold and has increasingly become an exurban, Republican leaning county. I’m projecting a 6-8 point Sanders win here. In my scenario, this is the one state Sanders wins on Tuesday night, though like I said, I’m almost 50/50 on whether he’ll also win Ohio narrowly.

Florida (214 delegates):

Here’s the thing. Florida is a closed primary for one, and that means none of the independents that have boosted Sanders can vote in it. Secondly, let’s just say the South has been a series of Michigans, but Hillary’s Michigans. She has consistently overperformed her polling everywhere in the south. Here is a rundown:

Louisiana: (Two polls averaged) Clinton at 60.5%. Election day total: 71.1%. Polling average: -10.6

Mississippi: (Two old polls) Clinton at 62.5%. Election day total: 82.6%. Polling average -20.1

Arkansas: (Again, only two relevant, slightly old polls): Clinton 57%. Election day total 66.1%. Polling average -9.1

Alabama: (Just going to go with the one poll right before): Clinton 71%. Election day total:  77.8%, Polling average -6.8

Tennessee: (3 relevant polls) Clinton 57.3%. Election day total: 66.1% Polling average: -8.8

Georgia: (7! polls this time): Clinton 63%. Election day total: 71.1%. Polling average -8.1

South Carolina: The final Pollster.com average was Clinton 58.4%. Election day total: 73.4%. Polling average -15

Virginia: (Average of final 5 polls): Clinton 56.4%. Election day total: 64.3%. Polling average -7.9

Texas: (Last 6 pollsters not including ARG!) Clinton 62.3%. Election day total: 65.2%. Polling average -2.9

The average polling underestimation for Clinton so far, in southern states, is: -9.9. Now this number is problematic because it weights equally for several states that only had 2 or fewer polls, in two occasions well before the actual election and so missed movement. Texas, South Carolina and Virginia were the most polled. Just messing around to find a simple (and not precise) weighting system SC x3, VA x2, and Georgia x3, Texas x4, Louisiana, Mississippi and Alabama at 1/2 might help put the misses in perspective, but it still equals an approximately 7.5% average miss. In fact, just looking at the well-polled states, Georgia, South Carolina, Virginia, and Texas, we see an average underestimation of 8.5 for Clinton’s final percentage. That is very bad news for Sanders in Florida, where she is averaging a 60.9-31.5 lead.

I would say more so than polling being very bad in the south, it’s just that undecideds have universally broken for Clinton there, and heavy. Now Sanders is making a casual attempt to keep Florida from being too much a blow out, it most resembles Texas in many ways. So I think Hillary will have a Texas style over-performance, which is to say much more moderate and which presumes undecideds are split roughly 50-50. 3.5% over what polls have her at, for a 64.4% win.

What’s bad is that Sanders is still at only 86% of where he needs to be to tie Clinton in delegates in 538’s rather extensive model, and even with the Michigan upset, he fell further behind where he needed to be at this point in the race. A 64-36 loss in Florida, is not good for Sanders. Florida has the same number of delegates as Missouri and Ohio combined. A loss like that could put him further behind in where he needs to delegate-wise, even with wins in Ohio and Missouri. A similar split in Texas gave Hillary a 70 delegate margin, and with Florida a 64-36 win should be enough for about a 62 delegate margin depending on congressional district splits. Sanders need Florida to be -18 delegates in the 538 split. A 44 delegate overperformance there by Hillary, would be enough by itself to undo massive Sanders upsets and overperformances in Illinois, Ohio, and Missouri.

North Carolina (107 delegates):

All the stuff about the south goes equally for North Carolina. The last two polls have at 57% and 58%, with Elon being junk, and there not being a lot of recent polling. Looking at the states bordering NC, Tennessee gave Hillary 66%, South Carolina 73%, and Virginia 64%. I’d imagine North Carolina will be within the lower end of that range. I’m expecting between 60-65% or Hillary there, splitting the difference, 62%. To be on target, Sanders needs a 57/50 split in North Carolina, and it seems highly unlikely that he will get that, which would mean additional underperformance in the south.

But at least with this next week, Sanders can breath a sigh of relief at having made it through the south, finally. Of course there are still Maryland, Delaware, and Washington D.C. which all strongly favor Hillary Clinton and give her opportunities to overperform in this block of 136 delegates. Clinton also has, arguably, the advantage in other later states like Pennsylvania (like Michigan but closed primary and easily backed her in 2008), Puerto Rico, Rhode Island, New York (closed primary, very strong Democratic machines and union presence, quasi-home state advantage) and New Jersey. So by no means does the rest of the calendar after this point just favor Sanders. Even California, the really giant delegate prize, doesn’t offer large innate advantages for either candidate, and neither side has much if any, chance of winning it by a large margin.

And if Clinton sweeps all of the states this Tuesday and Sanders supporters try to claim the primary is still ongoing because neither candidate is even close to the 2300 mark. Well, that’s about the same as Karl Rove exploding on Fox news in 2012, or a Republican in 2012 saying “Only 20% of the votes on the west coast have been counted. We could sweep those 3 states and win the election. It’s too early to call this!”