I promised a weekly column covering analysis of the polling data, and then last Friday I didn’t deliver. I was on vacation in Portland, and tried to write a piece on my iPad on the plane and it just didn’t work, lol. We will survive. If anyone wants to push this up into community spotlight, I’ll write it for general weekend reading.
We have a lot to cover, so LET’S HIT IT. Math...exciting. Lol.
Bottom Lines (at the top!)
This week is reinforcing the narrative that the “Kavanaugh bump” the GOP enjoyed was temporary. If there ever was a bump at all...it could have just been normal polling fluctuations, blown out of proportion by a desire for a news story. Whatever was going on, polls have generally reverted back into blue wave territory, which we will discuss.
A second major polling story this week is the apparent divergence of the House and Senate projections, with it looking more and more like a Dem landslide on the House side, and a modest GOP victory in the Senate. I’ll discuss this in the Senate section, but yes...this appears to be possible.
Thirdly, early voting has started! We seem to be at the very early stages of an electoral narrative where this looks like a high turnout election. So I’ll add some early voting stats to this report, which aren’t polls per se.
Early Voting is Smashing Records
A high turnout election is generally seen as favoring us, and the reports that are coming in make this look like an election where turnout will look more like 2016 (55.4% turnout) than 2014 (36.4%). That is, the early voting reports I cite below are all more than doubling 2014 at this point, while just falling short of 2016.
Early voting and absentee ballots tend to be the sort of thing where we might be inclined to pull out the Tarot cards or chicken bones. The data is never clear because we can’t be entirely sure it means. Several viable scenarios present themselves in the 2018 data that shows a spike in early voting:
1. Dems are utterly enthusiastic, the bump is mostly our voters, and this is a sign of an impending blue tsunami.
2. Dems are pretty enthusiastic, and the bump is largely our voters, but these are largely people who would have voted regardless, who want to get their vote in early. This might signify a blue wave, but not an epic tsunami.
3. Both sides of the electorate are enthusiastic, so that there will be increased turnout, but an utterly unpredictable mix that we won’t understand the nature of until vote counts start coming in.
4. Early and mail voting has been increasing over the years regardless, so perhaps this data doesn’t mean much, it’s just the normal turnout being spread out over a longer period.
When compared to all of the other metrics — the generic poll, fundraising, district polls — I’m inclined to lean toward scenarios one and two. But we should underline that we just don’t know what this means yet.
Some specific results:
Georgia: the news out of Georgia was impressive enough to make Rachel Maddow, so let’s start there. First day voting, including mail ballots, totaled 129,458, vs. 46,086 in 2014. Early voting being 281% of the last midterm...that’s notable. And GeorgiaVotes.com shows the trend continuing as early voting progresses. Totals are sitting around 294% of 2014 levels.
Tennessee: Nashville county reports that 10,249 people voted on the first day, close to 2016’s 12,300. Knox County also reports record turnout that is slightly above 2016 levels. Statewide Reporting puts state totals at 120,893 vs 143,141 in 2016. Early voting was different in the state in 2014, so we don’t have a midterm comparison, but this is another metric that looks like an election where turnout will be closer to 2016.
Virginia: I can’t find statewide numbers, but local papers are reporting a 90% increase in early voting statewide (come on guys, that’s really bad writing to not include state totals or cited sources). Given the blue wave in the state earlier in the year, I’m going to suggest that the spike is good news.
Minnesota: The state is actually reporting increased early voting over 2016, with 301,107 ballots requested, compared to 281,532 in 2016. We can’t use 2014 data because the law changed after that election so that people could early vote without an excuse. And that’s a caveat with MN early voting — voters might just be acclimating to the ability to vote by mail. California also saw huge increases in mail voting in the early cycles where we allowed anyone to vote by mail.
Indiana: Likewise, counties are reporting dramatic spikes in turnout for early voting.
Ohio: News out of the state is mixed, and I suspect partisan spin. What seems to be the case is that requests for early voting ballots have spiked — with 910k requested, vs 740k in 2014. But fewer have been received back by the registrar — 41k vs 49k in 2014. I don’t think we make anything of this data...I think total requests are more important than how many have been mailed in, but who knows.
That’s the early voting news I can find at this point. So let’s move on to polling.
Polls and Data
Broad Metrics
Trump approval: Currently at 42.2 (-10.2). Honestly, I’m not going to speak to this unless we go flying over 45% or under 37%. It’s pretty unremarkable where it is...this level of support could provide a president a modest win or a landslide defeat in the midterms.
Fundraising: As a metric, this data point makes this election look like an utter landslide. Consider this — in House races, Dems have taken 65% of the total funds. Wow. This is unprecedented. And we’ve seen statistic after statistic come in that speaks to the money game — 73 Dem House candidates have raised over $2 million, compared to 17 Reps. 144 Dems have raised over $1 million, compared to 84 Reps.
But money isn’t votes. I look at this data with both pro and con caveats. Pro — all of this money should allow Dems to fight these last weeks hard, and that could create some cross-race benefits as media markets get saturated with our message. Con — traditionally money correlates to turnout. But modern fundraising might change this, and donations might be more prone to come from outside the district, so that money raised in Kansas is a sign that Californians are enthusiastic...which doesn’t generate votes, obviously.
Generic Congressional Ballot Poll: 538’s aggregate sits at 8.4 points, up from 7.7 points at the beginning of October. Here’s what I think is the big news in the CGBP this week — pollsters are moving from registered voter models to likely voter models, and it isn’t hurting Dems at all. Usually this shift benefits the GOP by a few points, since their voters turn out better. But what we are seeing indicates that this advantage has been erased...maybe even turned to our advantage.
But let’s cover some of the notable pollster’s results.
Ipsos/Reuters: They’ve put out polls at D+12, +13, +11 and +10 in the last week. This is all consistent with a wave, and up from their +7 results that fueled the Kavanaugh bump narrative. They’re using likely voter models...these are great polls for us.
ABC/WaPo: This is the gold standard A+ pollster, and they’re saying D+13. This matches their last poll, which was the end of August and came in D+14. But note — they switched from registered to likely voters between the two. Great news for Dems. CNN/SSRS, another A-rated pollster, reported D+13 in an LV result.
Rasmussen: Sign. Effing Rasmussen. Mostly it bothers me because I have to explain my theory every time, in case there are new readers. I strongly believe that they purposefully use models favorable to Republicans on polls that make for good propaganda. They returned a tie and D+1 in two recent polls. This makes me think they’re using a 2014 electorate model. Could they be right? Sure. That electorate could turn out, but I think what’s more likely is that they’re wrong and they’re holding down the aggregates because of it.
YouGov: This pollster has consistently been more bearish on 2018, and reported D+5. They’ve been reporting D+4 to D+7 for months...they’re probably good polls, but I have no idea how they’re getting results that are so flat.
KFF: They return D+12. The poll is worth reading — lots of healthcare related polling in here, with the bottom line being a confirmation that this is a top issue in 2018...probably the top issue.
Overall: We’ve returned to a relative high point in this metric’s cycle, with over a third of the polls coming in at double-digit Dem leads. As always, we have to be cautious with this metric, because a D+10 wave resulting from four million Los Angeles County Dem votes isn’t that helpful, where D+6 with high turnout in competitive districts could still be quite impactful.
The big question:
As Nate Silver points out today, pollsters have generally agreed on a likely voter model that assumed roughly equal turnout between Dems and Reps. The special elections showed higher Dem turnout. Pollsters have made an educated guess at turnout, but if they are off by a couple of points in either direction we might see a big shift in outcome vs prediction.
Senate
A major narrative emerging is the potential for the House and Senate to go in opposite directions. 538 give Dems greater than 80% odds at capturing the House, but less than 20% odds of winning a Senate majority.
I’m not clear why this is surprising or noteworthy. The types of races we’re facing on the Senate side are very different than the average House race. It’s a red state-heavy map. If the competitive House districts all looked like Tennessee and North Dakota, we’d be having a rough time on the House side too. But the ample targets to the left of this sort of districts mask the fact that the blue wave can’t erase sheer GOP dominance in an area.
So the Senate remains a tough map, and this week’s polls reinforce that.
TN: A NY Times / Sienna live poll returned a disastrous R+14 result, fueling a lot of doom and gloom in this race. Ipsos/Reuters returned R+3, not as bad, but it’s clear that TN needs work if we are going to win it. SSRS returned D+1 on Thursday. The curve seems to indicate we can squeak it in a blue wave or better scenario.
TX: Tons of polling here, and it’s not great. Cruz is ahead +4 to +9 in all of them. But let’s remember...the current polling offers Beto similar odds to Trump in 2016. We haven’t lost this one yet. To me, a Beto win is a narrow path, but one that has predictable needs. We need young voters to turn out. We need hispanics to turn out. Both groups have histories of failing to do so.
But there is some good news in the clutter — Texas voter registration is breaking all the records. We don’t have final figures from the Oct. 9 deadline yet, but it appears the state will blow through most every registration metric. And the registrations seem to be coming from the right areas. Austin reported over 35,000 last day applications, compared to 10,000 in 2016. We could be looking at early evidence that young and hispanic voters are indeed voting.
And the GOP narrative in the RWNJ bubble is ‘har dee har, Dems thought they’d win Texas, but we are up 10 in every race and the Senate looks like a red wave.” I’m not sure that message is the best to prevent a surprise loss in this race and others.
NV: This pickup looks a bit better than the previous two. NYTimes/Siena came back with R+2, and Emerson says R+7. Vox Populi says D+2 and KFF R+1. But 538 still offers a decent 40% shot at this one. Nevada is pretty unpredictable with turnout, so this one is up in the air.
AZ: Weird polling here...NY Times has a live poll in progress, but I don’t quote incomplete polls. Check it — the lead has flipped like five times as the polling has progressed. Some random pollster called Data Orbital says D+7. 538 continues to rate this a decent shot at a pickup, 60%.
ND: This is our most likely loss. The polls have turned against Heitkamp, but no new polling this week so we don’t know if the Kavanaugh bump has subsided in this race. She remains an underdog by 538’s model, with a 33% shot.
MO, FL, IN, WV: I’ll lump these. No really significant polls here. Dems remain favorites in all, to varying degrees. The first two are barn burners, with poll results flipping between leaders in the +2 range, while the other two look like we are on track with Donnelly and Manchin generally up by high single digits.
House
Jeebus there’s a lot of polls, lol. And a lot of partisan garbage polls. I’m going to try to hit the highlights, and quote ONLY completed polls from reputable pollsters. Campaign internals don’t tell us much.
Overall the House is looking great. 538 gives us an 83% chance of taking the chamber, with an average seat gain of 40. CNN’s Harry Enten is using a larger margin of error, and his average seat gain is currently 34.
PA-01: Siena/NY Times returned D+7 in this Wallace vs. Fitzpatrick matchup. It raised our odds in the district from 20% to 50% per 538.
NJ-07: This is an interesting race/poll with an R+1 result for GOP incumbent Lance, with Dem Malinowski still rated a 2-in-3 favorite by 538. This cycle is insane. It feels like there are fifty House races where we could conduct two simultaneous polls and get different projected winners.
Interesting — these are the only Cook toss-up races that were polled this week. Hmm. I’d have thought they’d be of high interest. Guess not.
WV-03: Monmouth says R+2 in this Miller vs Ojeda race. What’s interesting is that the gap is huge in the 538 models when comparing polls-only to expert models (both show a probable GOP win). This race suggests to me that some districts are seeing moderately significant realignments...but the election will be the test of this, and the measure of whether it will be enough for the candidates.
FL-15: Carlson Vs Spano. The Siena poll here in in progress, they need 70-100 more responses. I cite it because this is a ‘lean R’ district on Cook’s list, and it looks like this will be a <2 point margin in either direction. SurveyUSA also polls...and got a dead heat. Likewise, Siena is polling VA-05. It’s too early, but this is a Cook Lean R district and the results are currently a dead heat.
House notes: I’m disappointed in the available polls. I audited Cook’s entire list of Toss Up and Lean R races, and above is pretty much all we got returned this week in polls. I don’t know...it makes it pretty hard to ascertain what’s going on. There are some really great polls for us, and some bad. But when the most important races don’t have new polling, it’s too much guesswork.
Governors
538 launched their Governors forecast, so anyone interested can get all the pertinent polling for each race there. The model suggests that Florida and Wisconsin are leaning our way, with Georgia, Nevada and Ohio sitting at toss-up status.
So How’s the Wave?
This part of the cycle always drives me crazy. I remember in 2014 that Republicans looked great in September, then we hit this bump in October that suggested Dems would counteract the wave. But then...nope...we got smacked.
The midterms feel like they are still teetering right on the dividing line between great and meh. If every race on the map shifted two points left by Election Day, it’d feel like a true wave scenario. But if things shift two points right, it could create a pretty mediocre result.
Honestly, I don’t know how we should interpret this week’s data. What I’d say is that many of the metrics are there that we’d want to see for a wave to come crashing in, but many of them are murky. Everything hinges on turnout. Are these bumps in early voting the resistance showing up to smack the GOP, or are both sides highly engaged? Will women really turn on the GOP? Will young voters show up?
It feels like the election isn’t settled yet. And honestly, that’s probably a better feeling for our side than assuming we have it in the bag. We are down to the wire, and we have an amazing slate of candidates, a pile of cash, and we are winning on the issues voters care about. If I had to pick which party’s position to be in, clearly it would be ours, and maybe that’s the best we can hope for.