Nate Silver's Blog, page 78
October 26, 2018
Politics Podcast: No, The Model Doesn’t Care About The Early Vote
More: Apple Podcasts |
ESPN App |
RSS
| Embed
Embed Code
Early voting has begun! In this installment of Model Talk on the FiveThirtyEight Politics podcast, Nate Silver explains why he doesn’t use early voting data to help forecast elections: Polls are more reliable, given that we don’t know all of the factors involved in early voting. Nate also answers questions from listeners, including which House race he would most like to know the results of if he could only know one before Election Day.
You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .
The FiveThirtyEight Politics podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.
Democrats Have A Whac-A-Mole Problem In The Senate
A lot of the midterm outlook simply comes down to exposure: which party has more of it, and in which chamber.
In the House, there are 111 (!!) competitive seats, according to the Classic version of our model – races where each party has at least a 5 percent chance of winning. Because the national climate favors Democrats and because Republicans control most of the seats in swing districts as a result of their strong performance in the 2014 and 2016 elections, the overwhelming majority of those are currently held by Republicans. Granted, some seats from among that list of 111 are fringy pickup opportunities for Democrats – but Republicans are playing a lot of defense in a lot of different types of districts.
Just the opposite is true in the Senate, where 26 of the 35 seats up for election this year are currently held by Democrats, including 10 Democratic incumbents running in states won by President Trump. With that kind of map, Democrats would have to catch almost every possible break to win the Senate, and that’s not usually how politics works.
In fact, Democrats’ odds continue to become longer in the Senate, with their chances down to about 1 in 6 in all three versions of our forecast. This is despite the fact that their position has become better in some individual races, as you can see in the chart below, which compares our current Senate forecast against the last time we did a version of this exercise two weeks ago.
Where Senate odds have changed
Democratic chances of winning on Oct. 9 vs. Oct. 26 (as of 8 a.m.), according to the FiveThirtyEight 2018 Senate forecast
Lite forecast (polls only)
Classic forecast (polls fundamentals)
STATE
9-Oct
26-Oct
CHANGE
9-Oct
26-Oct
CHANGE
Nevada
52%
38%
-14
50%
40%
-10
Tennessee
41
28
-13
29
24
-5
Mississippi special
23
17
-6
14
12
-2
Texas
23
18
-6
25
21
-4
North Dakota
14
9
-5
31
32
1
Arizona
68
64
-4
65
62
-3
New Jersey
88
87
-1
92
90
-2
Indiana
68
69
1
75
71
-4
West Virginia
88
89
1
88
89
1
Missouri
51
52
1
57
58
1
Minnesota special
85
87
2
89
90
1
Montana
79
83
4
84
86
2
Florida
57
75
18
60
74
14
Overall Senate
21
16
-5
21
17
-3
Races in which Democrats’ chances have consistently been greater than 95 percent or less than 5 percent in the Classic version of the forecast are not listed.
The Democratic outlook has brightened in Florida, for example, where incumbent Bill Nelson now has a lead in most polls and is roughly a 3-1 favorite to defeat Republican Gov. Rick Scott. It’s also gotten better in Montana, where nonpartisan polls have been relatively rare, but have continued to show Jon Tester ahead.
But other races look like bigger challenges for Democrats than before. In Nevada, Republican incumbent Dean Heller – the only GOP senator on the ballot in a state carried by Hillary Clinton — has pulled into a slight lead in the polls against Jacky Rosen. Although the polling has been mixed in Tennessee, most of it has gone against Democrat Phil Bredesen. A pickup in the Mississippi special election, always a bit of a long-shot for Democrats, now looks like more of a pipe-dream after recent polling there.
I realize that “some states are moving in one direction and some in the other direction” isn’t a great headline, but this nicely illustrates Democrats’ Whac-a-Mole problem in the Senate: Whenever they can start to feel a little better about their prospects in one state (Florida), another problem crops up (Nevada).
Nor is it really as though Democrats can curse their luck in the Senate. Sure, they’ve had a few bad breaks:
Scott entered the race and put up a vigorous challenge against Nelson, even if it now looks more likely than not to fall short.
Bob Menendez’s corruption trial has turned New Jersey into a semi-competitive race.
Al Franken’s retirement created an outside (and perhaps slightly overlooked) chance for Republicans to pick up a seat in Minnesota.
In North Dakota, Democratic incumbent Heidi Heitkamp has had a lot of self-inflicted wounds.
John McCain died after the deadline for a special election to be held this year in Arizona, which would have given Democrats a second opportunity to pick up a seat in that state.
But there have been roughly as many favorable or “lucky” developments for Democrats:
Bredesen and Beto O’Rourke have created legitimate opportunities for Democrats in Tennessee and Texas when those races didn’t necessarily look likely to be competitive at all.
Republicans nominated underwhelming candidates in several Midwestern states such as Pennsylvania, Ohio and Wisconsin.
The only Democratic retirement was Franken.
Republican Thad Cochran’s retirement created a special election and an outside pickup opportunity for Democrats in Mississippi.
And let’s not forget: the Democrats somehow won a Senate race in Alabama last year , thanks to a good candidate in Doug Jones and Republicans nominating Roy Moore.
It’s too soon to say the Senate is a lost cause for Democrats. A 1-in-6 chance isn’t good, but it isn’t zero, either. The scenario for a Democratic victory, however, probably involves a systematic polling error in which all of the polls are off in the same direction rather than Democrats clawing their way to victory on a race-by-race basis. The map was tough enough for Democrats that almost everything possible was going to have to go right for them to win the Senate, and it hasn’t.
October 25, 2018
Election Update: Romney-Clinton Districts Are Overrated. Obama-Trump Districts Are Underrated.
This year’s midterm hasn’t really featured the “model wars” we saw in 2014 or 2016 — heated arguments between different election forecasters, whose projections sometimes showed very different results. Instead, the various election models and forecasts have largely told the same story, one where Democrats are solid-but-not-certain favorites to win the House, and the same is true for Republicans in the Senate.
But there have been some differences of opinion, and one of them concerns what sorts of districts Democrats are most likely to pick up in the House. Under one theory, the midterms are mostly a referendum on President Trump. This theory posits that Democrats are likely to do especially well in districts — often racially diverse and urban or suburban — that have shifted toward Democrats recently, such as the 13 districts that voted for Mitt Romney in 2012 but Hillary Clinton in 2016. Conversely, Democrats might struggle to make gains, the theory would say, in the largely rural and white districts that have shifted the most toward Republicans recently — like the 21 districts that voted for Barack Obama in 2012 but Trump in 2016.
A contrary theory is that midterms are mostly about reversion to the mean and about voters trying to balance power between the two political parties. Under this theory, districts that have recently swung toward Trump but were purple enough to vote for Obama in 2012 might be trouble spots for Republicans. That’s because they tend to be “swingy” (or elastic) districts, and they don’t necessarily have a lot of long-term loyalty to the Republican Party. Moreover, to the extent Trump won these states and districts on the basis of marginal, late-deciding voters — something that was true in the Midwest, which is home to many of the Obama-Trump districts — those voters might not be as reliable for Republicans in 2018. Sometimes the most recent voters to enter your coalition are the first ones to head right back out.
These theories aren’t entirely mutually exclusive. The House playing field is quite broad, with Democrats having opportunities in all different types of Republican-held districts; that’s part of what makes their opportunity for gaining a majority more robust. Nonetheless, the Romney-Clinton districts have been the subject of an awful lot of attention and even talk of a “suburban tsunami”; the Obama-Trump districts, less so.
The thing is, though, if you actually look at the polls … Democrats are doing just as well in the Obama-Trump districts. Probably a little better, in fact. First, here are FiveThirtyEight’s current adjusted polling averages in the Romney-Clinton districts, all of which have at least one poll:
Who’s ahead in Romney-Clinton districts?
Polling averages as of 12 p.m. on Oct. 24
Presidential Result In:
District
Incumbent
Polling Average
2016
2012
AZ-2
—
D+11.2
D+4.9
R+1.5
CA-25
R
0
D+6.7
R+1.8
CA-39
—
R+0.7
D+8.6
R+3.7
CA-45
R
D+1.2
D+5.4
R+11.8
CA-48
R
R+1.7
D+1.7
R+11.7
CA-49
—
D+13.3
D+7.5
R+6.5
IL-6
R
R+1.0
D+7.0
R+8.2
KS-3
R
D+7.5
D+1.2
R+9.5
NJ-7
R
D+1.4
D+1.1
R+6.2
TX-7
R
R+4.0
D+1.4
R+21.3
TX-23
R
R+15.9
D+3.4
R+2.6
TX-32
R
D+0.3
D+1.9
R+15.5
VA-10
R
D+7.7
D+10.0
R+1.1
Average
D+1.5
D+4.7
R+7.8
On average, Democrats lead by 1.5 percentage points in polls of these districts. That’s pretty good given that most of them feature Republican incumbents, among whom only Rep. Will Hurd in Texas 23 has a decisive lead in the polls. But it’s also not spectacular. Clinton won these districts by 4.7 percentage points on average — more than Democratic Congressional candidates are leading them by (although Obama lost them by an average of 7.8 points).
And here are the Obama-Trump districts … and the Obama-Trump states! Trump won six states that Obama had carried in 2012, five of which — Iowa is the exception — feature a U.S. Senate race this year.1
Who’s ahead in Obama-Trump districts (and states)?
Polling averages as of 12 p.m. on Oct. 24
Presidential Result In:
District
Incumbent
Polling Average
2016
2012
IA-1
R
D+9.6
R+3.5
D+13.7
IA-2
D
D+16.6
R+4.1
D+13.1
IA-3
R
R+1.3
R+3.5
D+4.2
IL-12
R
R+6.5
R+14.8
D+1.6
IL-17
D
*D+30.0
R+0.7
D+17.0
ME-2
R
D+0.9
R+10.3
D+8.6
MN-1
—
R+0.4
R+14.9
D+1.4
MN-2
R
D+6.0
R+1.2
D+0.1
MN-8
—
R+11.8
R+15.6
D+5.5
NH-1
—
D+8.2
R+1.6
D+1.6
NJ-2
—
D+25.0
R+4.6
D+8.1
NJ-3
R
D+1.7
R+6.2
D+4.6
NV-3
—
D+1.4
R+1.0
D+0.8
NY-1
R
R+6.3
R+12.3
D+0.5
NY-2
R
*R-7.8
R+9.1
D+4.4
NY-11
R
R+10.2
R+9.8
D+4.3
NY-18
D
*D+26.5
R+1.9
D+4.3
NY-19
R
0.0
R+6.8
D+6.2
NY-21
R
R+5.9
R+13.9
D+6.1
PA-8
D
D+12.8
R+9.6
D+11.9
WI-3
D
*D+27.0
R+4.5
D+11.0
Average
D+5.5
R+7.1
D+6.1
Presidential Result In:
District
Incumbent
Polling Average
2016
2012
FL-Sen
D
D+3.4
R+1.2
D+0.9
MI-Sen
D
D+16.7
R+0.2
D+9.5
OH-Sen
D
D+13.7
R+8.1
D+3.0
PA-Sen
D
D+15.8
R+0.7
D+5.4
WI-Sen
D
D+13.1
R+0.8
D+6.9
Average
D+12.5
R+2.2
D+5.1
* There’s no polling in these districts, so polling results are imputed based on our CANTOR system, which looks at polls of similar districts to infer what polling would look like in districts that don’t have any polling.
As you can see in the chart, four Obama-Trump congressional districts don’t have any polling — perhaps a sign of how they’ve received less attention from the media this year. So in those districts, I’ve used our CANTOR system, which estimates what the polls would be in those districts if we had them, based on polling in similar districts. (It’s not a perfect compromise, but three of the four districts rate as “Solid D” according to both FiveThirtyEight and the Cook Political Report, and would likely show large Democratic leads if they’d been polled, so it would bias the sample to exclude them.)
This is an eclectic mix of districts, including several that are slam dunks for Democratic incumbents and others where Republicans are pretty clear favorites. But on average, Democrats lead by 5.5 percentage points in the polls there (counting the CANTOR districts for which we’ve used imputed polls). That roughly matches Obama’s 6.1-percentage-point win in those districts in 2012 and is much better than Clinton’s 7.1-point loss. Excluding the CANTOR districts, Democrats’ average polling lead is 2.3 points in these districts.
Meanwhile, Democrats are doing extremely well in the five Obama-Trump states with Senate races this year. On average, their candidates lead by 12.5 percentage points, and only Florida’s Bill Nelson is in any real danger of losing. These candidates are all incumbents, and you’d expect incumbents to beat the partisan lean of their districts, but Obama was also an incumbent in 2012, and he won these states by an average of only 5.1 percentage points.
So for my money, Democratic performance is a little better overall in the Obama-Trump districts than in the Romney-Clinton districts. You could credibly argue that it’s about the same if you account for the fact that there are more Democratic incumbents in the former group than the latter group. But at a minimum, the polls haven’t really matched the media narrative about where Democrats are performing well.
For every Romney-Clinton district where Democrats are excelling in the polls, such as Kansas 3 or Virginia 10, there are others — Illinois 6, say, or several of the districts in California — where they have middling numbers. And while there have been a few Obama-Trump districts where Democrats’ numbers have been poor lately — particularly Minnesota 8, in Minnesota’s Iron Range — those are more the exceptions than the rule. More typical is Maine 2, a Trumpy, swingy, white, secular, rural district, where Democrat Jared Golden has moved into a tie or perhaps a slight lead in the polls against Republican incumbent Bruce Poliquin.
Another way to approach this issue is to see how correlated the polls are with presidential results from 2016 and 2012, respectively. And that’s what I’ve done in the final chart, below. I’ve run correlation coefficients between the FiveThirtyEight polling average in every House and Senate race with polling and their results in the prior two presidential elections. The correlations are weighted based on how much polling the states and districts have, so places with more recent and more robust polling have more say in the calculation.
Congressional polls look (a bit) more like 2012 than 2016
Correlation between 2018 polling and 2016 and 2012 polling, respectively, in different kinds of races
Correlation with:
Races w/Dem. incumbents
Races w/GOP incumbents
Open-seat races
All races
2016 presidential results
+0.60
+0.69
+0.57
+0.61
2012 presidential results
+0.70
+0.64
+0.66
+0.74
Correlations are weighted based on the volume and recency of polling in each state or district.
Polls of this year’s Congressional races are actually more correlated with 2012 presidential results (a correlation coefficient of .74) than with 2016’s (.61). The effect is somewhat reduced if you account for the incumbency status of the districts. But based on the polling, the map looks at least as much like 2012 as 2016. Maybe the midterms elections aren’t all about Trump after all.
Not coincidentally, this was also the pattern in the various special elections around the country this year and last year: Their results were more correlated with 2012 than 2016. For instance, probably the worst result among the special elections for Democrats was Jon Ossoff in Georgia 6, the wealthy district in the Atlanta suburbs that voted for Romney by a wide margin in 2012 but nearly voted for Clinton in 2016. Democrats have multiple paths to win the House — but if they do put together a majority, don’t be surprised if their route there involves more Maine 2’s and fewer Georgia 6’s.
October 24, 2018
We’re Back From The Future. Which Of These Wildly Different Midterm Outcomes Would You Believe?
Welcome to FiveThirtyEight’s weekly politics chat. The transcript below has been lightly edited.
natesilver (Nate Silver, editor in chief): Today I’m coming to you from the future. Or maybe I’m not. The rest of you will have to figure out whether I’m really coming from the future or just pulling your leg. But for now, I’ve timeported myself back from Nov. 7, and I’m going to tell you what happened on Election Day. Tell me whether you believe me or not.
nrakich (Nathaniel Rakich, elections analyst): There’s approximately a 0 in 1 chance that Nate is coming to us from the future.
natesilver: Here’s what happened: Democrats picked up only 16 House seats, short of the 23 they needed to take control of the House. They might win another one or two once California finishes counting its returns, but they’re not going to get to 23.
And in the Senate — oh, boy.
Democratic incumbents lost in North Dakota, Missouri and Indiana. And they didn’t gain any Republican seats, although there’s a chance that Kyrsten Sinema could win Arizona if it goes to a recount (don’t get your hopes up, Democrats — she’d need almost everything to go right). So it looks like we’ll end up with a 54-46 Republican-controlled Senate.
There were a few bright spots for Democrats. Andrew Gillum won the Florida gubernatorial election. Republican Gov. Scott Walker lost by 2 points in Wisconsin. But overall, Republicans beat their polls by an average of 2 to 3 points — a lot like 2016, to be honest.
Democratic turnout was high. But GOP turnout was high too, and the election was fought in very red places.
OK, so that’s what happened.
Why don’t you guys ask me some questions so that you can verify I’m telling the truth?
julia_azari (Julia Azari, political science professor at Marquette University and FiveThirtyEight contributor): I’m not sure what kinds of questions. I think I don’t understand the rules of this game.
But this scenario seems plausible to me.
What was wrong with the polls in Arizona?
natesilver: Not sure you could say much was wrong with them — Sinema was up only 2 points in our forecast, and she’s down by 0.3 points based on votes counted so far.
That’s the thing — the polls weren’t really that far off. Just almost all of them were off in the same direction.
nrakich: That scenario is unlikely in the House, but it’s believable in the Senate. A Democratic gain of 16 House seats falls outside the 80-percent interval of the most likely possible outcomes in our model. That would mean Democrats basically just won the seats that are clearly in their favor right now — e.g., the Iowa 1st, the Virginia 10th, a handful in Pennsylvania — but not much else.
There’s about a 20 percent chance that Republicans pick up three or more seats in the Senate. Joe Donnelly of Indiana and Claire McCaskill of Missouri would probably be two of the first Democratic incumbents to fall in that scenario.
natesilver: I think pollsters are still sorting it out, but there’s a sense that wavering Republican voters came home to Trump at the end. We’re in an era where everything is very polarized along presidential lines, and that little uptick in Trump’s approval rating at the end may have helped, even though we were dismissive about it beforehand.
julia_azari: This is my sense about 2016 and politics in general: Predictability is making things, weirdly, slightly harder to predict.
sarahf (Sarah Frostenson, politics editor): I think there’s something fishy in Future Nate’s count of House seats that Democrats picked up.
What can you tell us about those House seats?
natesilver: They didn’t necessarily fit any one pattern. GOP incumbents were a bit stronger than expected in some places. And Democrats were disappointed with the results in Texas and Florida, where they didn’t get much of a Hispanic turnout. There were some impressive Democratic victories — Richard Ojeda somehow won in the West Virginia 3rd, for example, which just doesn’t make a ton of sense given how the rest of the night went, but that’s what happened. There just weren’t enough of them.
nrakich: Here’s a question for you, Nate: What was Trump’s approval rating on Election Day?
natesilver: The national exit poll had it at 45 percent.
But, again — Democrats actually had a fairly good night, in some ways! They won the popular vote for the House by 6 points. They just lost nine battleground races by 2 percentage points or less.
julia_azari: What were the implications of this for the “shy Trump voter” phenomenon? Does this result debunk it, since it happened when Trump wasn’t actually on the ballot? Is the problem with the sampling or the likely voter screen? Or just unavoidable error? I want to know what this means for the nationalization of politics.
Also, were there any purple states where voters split the ticket on governor and Senate races?
natesilver: Sure. Ohio split, for example.
Plus some of the obvious ones, like Massachusetts and Maryland.
julia_azari: Right, Massachusetts is weird.
nrakich: Hey.
julia_azari: Did any of the top-shelf competitive races split? Did Sen. Tammy Baldwin win?
natesilver: Of course Baldwin won — it was always a Republican pipe dream to beat her.
julia_azari: I’ve seen a lot of yard signs for Leah Vukmir, her Republican opponent … (Please note that I am trolling Nate and this is not a factual statement.)
What about Bill Nelson? Democrats held on to his Florida Senate seat?
And wait … none of the competitive states that had both a governor and Senate race split? Just blue states with popular red governors?
natesilver: Yeah, Nelson held on.
And yet again, yard signs weren’t that predictive. But to your earlier question, Julia — there is some question about a “shy Trump” effect. The random digit dial polls tended to show better results for Democrats than the list-based sampling polls, and they may have been more affected by Democrats’ superior enthusiasm.
nrakich: So both Nelson and Gillum won in Florida, but Democrats were disappointed with low Hispanic turnout?
The thing about Florida is that it’s a state where Hispanic voters might help Republicans. Those southern Florida districts are heavily Cuban-American, a GOP-leaning group.
But honestly, that’s the only thing that seems inconsistent to me about your narrative. I buy that you’re from the future.
natesilver: Nathaniel, I think you’re neglecting that Nelson and Gillum both went into the election up several points. They were up by 4 points and 6 points, respectively, in our final polling average, and they each won by about 2 points.
nrakich: But why would Democrats be disappointed? They still won.
natesilver: Because they lost almost all the toss-up House races there, other than the Florida 27th (which maybe shouldn’t have been thought of as a toss-up to begin with). In the 26th, Republican Carlos Curbelo hung on — by 9 points, in fact.
julia_azari: So what we’re doing here is looking at a lot of small shifts in our expectations — no major upsets?
natesilver: What we’re doing here is that I’m coming back from the future and telling you what happened.
In the House, we had nine candidates (seven Republicans and two Democrats) win in races where they had less than a 20 percent chance of doing so. So there were some surprises there. Not as much in the Senate, really.
nrakich: Who was the other Democrat (other than Ojeda)?
natesilver: Ammar Campa-Najjar, of all people — bad night for Duncan Hunter in the California 50th.
Although, I should note that technically it’s still possible for Hunter to win — it’s just that the late mail ballots are expected to be Democratic.
sarahf: And what can you tell us about the seven Republicans? Any trends there?
natesilver: Yeah, there were several races where the Republicans were only slightly behind in the polls, but our model was very bullish on Democrats based on their fundraising.
The fundraising was a good sign of Democratic enthusiasm. But Republicans were enthusiastic too, even though they didn’t donate much money. Big-time donations from donors like Sheldon Adelson helped a lot. Republican efforts to triage basically worked.
nrakich: A lot of the fundraising was inefficiently distributed (*cough* Beto *cough*), so I certainly believe that the fundraising could fall flat.
julia_azari: I find how seemingly mundane the surprise is compelling, but my objection on the Senate side is that if things go badly for the Democrats, there will be at least one more incumbent loss — in Florida, or in Montana, or a true upset in Wisconsin, where underpolling of rural voters may be a real thing.
natesilver: Jon Tester won by 9 points in Montana. That was sort of a surprise in the opposite direction.
nrakich: Small state, popular incumbent. That checks out.
natesilver: But wait. Something’s happening. Something strange.
I have to go back into the timeporter.
[several hours pass]
julia_azari: I call BS on multi-directional surprises.
natesilver: Well, folks. I just got back from the future. It turns out that Democrats had a really, really good night. They’re going to win somewhere between 53 and 55 House seats, depending on a couple of California districts. And they somehow won the Senate, thanks to Texas! I guess Beto O’Rourke is going to be our next president now.
You guys should probably ask me a few questions about what happened, since it’s not too often you get to talk to someone from the future.
nrakich: Did Heidi Heitkamp win in North Dakota, or did Democrats have to make up for that loss elsewhere?
natesilver: She lost by 8.
But Beto O’Rourke won, and Democrats won all the toss-ups.
julia_azari: Where else did the Democrats pick up in the Senate?
What was the national House popular vote?
natesilver: Democrats won Arizona by 7 (!) and Nevada by 5 (!). Pollsters really underestimated Democratic performance in the Southwest.
The Democrats made huge gains in California, and it’s not out of the question that Republican Rep. Devin Nunes could lose once all the mail ballots are counted (although I wouldn’t bet on it).
nrakich: Eh, I don’t know about that. I’d say Democrats’ best path to a Senate majority lies with the incumbent in an underpolled, small state winning, not with a pro-impeachment Democrat who’s trailing in about a bazillion polls defeating Ted Cruz.
natesilver: But, Nathaniel, the polls weren’t that great.
For some reason, people don’t blame the polls unless Republicans beat their polls.
Democrats are going to win the popular vote for the House by something like 11 or 12 points, though.
sarahf: Was turnout lower among Republicans, too?
natesilver: It’s not that turnout was lower among Republicans. It’s just that all cylinders were firing for Democrats: There was very high turnout among suburban women and baby boomers and relatively high turnout among black/Hispanic/Asian voters. And independent voters went Democratic in the House by 14 points in the national exit poll.
julia_azari: To go back to an earlier question, what was Trump’s approval rating on Election Day?
natesilver: 44 percent.
But the small fraction of voters who had no opinion of Trump overwhelmingly backed Democrats.
nrakich: I’d bet more money on the polls in North Dakota being off than the polls in Texas. We have multiple high-quality pollsters in Texas. In North Dakota, it’s Fox News (a very good pollster) and then a handful of outlets without long track records of success.
natesilver: You can bet as much money as you want — but I’m from the future, and I know what happened.
Pollsters were baking in too many assumptions about a 2010- or 2014-type turnout — if they’d relied on self-reported voting intention, they would have come a lot closer to predicting the Democratic surge.
nrakich: That’s a narrative I would believe if Nate’s scenario came to pass.
julia_azari: What happened in the governors races?
natesilver: Democrats won all the highly competitive races, except in Kansas and Alaska.
Scott Walker actually almost held on in Wisconsin, losing by just 1 point, but a loss is still a loss.
nrakich: So Walker did better in your “Democratic surge” scenario than your “revenge of the Republicans” scenario? 
October 23, 2018
Trump’s Approval Rating Is Up. Republican House Chances Are Down. Does That Make Any Sense?
Just in time for the Nov. 6 election, President Trump is posting some of the best job approval numbers of his presidency. His approval rating is currently 43.1 percent, according to the FiveThirtyEight average, the highest it’s been since March 2017. His disapproval rating is 52.0 percent, so let’s not get too carried away — the president is still unpopular. But his numbers are moving in the right direction.
Republicans have been moving in the wrong direction in our House model, however. The three different versions of our model — Lite, Classic and Deluxe — each show slightly different results, but they all have Republicans at or near their all-time lows. GOP chances are down to 14 percent (about 1 chance in 7) in the Classic model, our default version. The Lite version, which is based on polls alone and doesn’t account for factors like fundraising, is slightly less bearish, putting Republican chances at 21 percent (about 2 in 9). But that’s still as low as they’ve been all cycle.
So, what gives?
The Classic and Deluxe versions of our model actually use presidential approval ratings in their calculation. But they’re a relatively minor factor — part of an initial projection the model makes based on historical data to calibrate its forecast early in the cycle. Now that Election Day is almost here, the historical projection plays little role.
That’s because presidential approval ratings have a relatively rough relationship with midterm outcomes. More direct measures of voting preferences for Congress are more reliable. Foir example, the generic congressional ballot, which shows a Democratic lead of 8.6 percentage points, is bad for Republicans and has been getting (slightly) worse. And the large array of House district polls from Siena College and our friends at The Upshot have reverted back to being quite poor for Republicans after a couple of weeks in early October when they showed more bright spots for the GOP.
Besides that, Republicans are doing about as well — or as badly — as you might expect given the long history of the president’s party performing poorly at the midterms and that Trump is at least somewhat unpopular.
In fact, let’s take Trump entirely out of the equation for a moment. On average in midterm elections since World War II, the president’s party has experienced a popular vote swing of 7.3 percentage points against it, relative to how it did in the House popular vote in the presidential year two years earlier. Republicans won the House popular vote by only 1.1 percentage points in 2016, for example, so if you knew nothing else other than that a Republican was president, you’d expect them to lose the popular vote by 6.2 percentage points this year. That’s only 2 or 3 percentage points better than their 8- or 9- point deficit on the generic ballot and in our forecast of the House popular vote this year.1
Nonetheless, Trump remains less popular than average. If it holds to Election Day, Trump’s current net approval rating, -8.9 percentage points, would be the fourth-worst out of 19 midterms since World War II, better only than Harry Truman’s in 1946, George W. Bush’s in 2006 and Barack Obama’s in 2014. (For all presidents, approval ratings are derived from our historical database.)
Presidents suffer in midterms — unpopular ones especially
Net swing in the presidential party’s House popular vote at midterm elections from previous election, 1946-2014
Presidential Party’s Popular Vote Margin in House
Year
President
pres. Net Approval Rating
Prev. elex
Midterm
Net Swing
1946
Truman
-19.0
+4.7
-8.5
-13.2
1950
Truman
-2.3
+7.2
+0.7
-6.5
1954
Eisenhower
+37.4
-0.5
-5.5
-5.0
1958
Eisenhower
+30.0
-2.5
-12.4
-9.9
1962
Kennedy
+36.7
+10.0
+5.3
-4.7
1966
Johnson
+2.7
+14.7
+2.7
-12.0
1970
Nixon
+29.0
-1.7
-8.7
-7.0
1974
Ford
+19.8
-5.6
-16.8
-11.2
1978
Carter
+13.1
+13.6
+8.9
-4.7
1982
Reagan
-6.1
-2.6
-11.8
-9.2
1986
Reagan
+34.0
-5.1
-9.9
-4.8
1990
H. W. Bush
+18.7
-7.7
-7.8
-0.1
1994
Clinton
+3.5
+5.0
-7.1
-12.1
1998
Clinton
+33.5
+0.0
-1.1
-1.1
2002
W. Bush
+32.7
+0.5
+4.8
+4.3
2006
W. Bush
-19.7
+2.6
-8.0
-10.6
2010
Obama
-4.3
+10.6
-6.8
-17.4
2014
Obama
-11.3
+1.2
-5.7
-6.9
2018
Trump
-9.0
-1.1
—
—
Average*
+12.7
+2.5
-4.9
-7.3
* Excludes 2018
Furthermore, the president’s party tends to do worse in the House popular vote than the president does in his approval rating. On average in midterms since 1946, the president has had a +11.3 percentage point net approval rating as of Election Day, but the president’s party has lost the popular vote for the House by 4.9 percentage points. That gap is partly explained by voters’ tendency to seek a balance of political power even if they like the president — something that may be especially relevant this year because, when voters voted Republican for Congress in 2016, most of them expected Hillary Clinton to be president. Republican candidates for the House won voters who disliked both Clinton and Trump by 30 percentage points in 2016. If, hypothetically, Democrats won that group of voters by 30 percentage points instead this year, they’d win the popular vote by around 10 percentage points, not far from where most generic ballot polls have the race.
There still is some relationship between a president’s approval rating and his party’s midterm performance. But even at Trump’s slightly improved values, it isn’t a relationship that bodes well for the GOP. The data in the chart below, for example, would imply roughly a 10-point popular vote loss for the GOP given Trump’s current approval rating:
Our model performs a slightly-more-sophisticated version of this calculation; for instance, it attempts to estimate Trump’s approval rating among likely voters, which is slightly better than his overall numbers.2 Still, it thinks Republicans “should” lose the popular vote for the U.S. House by about 8.3 percentage points — again, very close to what the generic ballot and other indicators show. And the modest improvement in Trump’s approval rating doesn’t help that much; the calculation would show Republicans losing by 8.8 percentage points if we’d used the Labor Day version of Trump’s approval rating instead of his current one.
The relationship between presidential approval and a party’s performance at the midterms has been stronger in recent elections. In the chart above, for example, the slope of the line would be almost twice as steep — meaning a clearer relationship between presidential approval and the House popular vote — if you’d only calibrated it based on midterm elections since 1994. But again, that isn’t a helpful set of facts for Trump. If we’d used only the most recent data, our model would show a worse projection for Republicans, because Trump is unpopular and unpopular presidents have been punished especially heavily by voters in recent midterms.
In short, while it might seem a bit weird that presidential approval ratings and the generic ballot have moved in opposite directions, the data isn’t that hard to explain. The president’s party usually does poorly at the midterms even with a popular president, and Trump isn’t popular. His numbers are improved, but only marginally. Moreover, the relationship between presidential approval and midterm performance is rough enough that you wouldn’t necessarily expect them to move in lockstep with one another. And Republicans aren’t doing any worse (or better) than you’d expect from historical trends. Maybe the Republican outlook in the House would be even worse without the recent uptick in Trump’s approval rating. But that outlook isn’t good, and while Trump is probably still a net liability for the GOP, Republicans have plenty of problems of their own making too.
October 22, 2018
Politics Podcast: What On Earth Should We Make Of The Democrats’ Huge Fundraising Edge?
More: Apple Podcasts |
ESPN App |
RSS
| Embed
Embed Code
House Democratic nominees are outraising Republicans in individual contributions by about two-to-one. The FiveThirtyEight Politics podcast crew debates how to make sense of a fundraising advantage that has no recent precedent: Does it foretell a better result for Democrats than expected, or have changes in fundraising methods made it a poor predictor of results? The team also reviews the marquee governors races on the ballot this fall.
You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .
The FiveThirtyEight Politics podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.
October 19, 2018
Election Update: Democrats’ Unprecedented Fundraising Edge Is Scary For Republicans … And Our Model
Yesterday, my colleague Nathaniel Rakich wrote about the Democrats’ impressive third-quarter fundraising haul, which boosted their odds in our forecast in a number of competitive House races. I’m going to hit the topic of fundraising once more, just to underscore how much of an outlier the Democratic advantage is relative to historical norms and how that could represent a challenge for our forecast.
It would be one thing if Democrats were raising money only in a few high-profile races — say, for example, in Rep. Beto O’Rourke’s Senate race in Texas. But that’s precisely not what is happening. Instead, the Democrats’ fundraising advantage is widespread. They’re raising money almost everywhere they need it in the House, whereas Republicans are sometimes coming up short.
For instance, we project that by the time they file their 12G reports later this month — the last filings due before the election — 144 Democrats on November House ballots1 will have raised at least $1 million in individual contributions, not counting self-funding or outside money. But we project just 84 Republicans will have done the same. We also project that 73 Democratic House candidates will have raised at least $2 million, as compared to just 17 Republicans.
Until recently, it was rare for House candidates to raise $2 million for their races, but it’s become more common in recent years as fundraising has gone digital and candidates have learned how to make highly tailored online appeals. There was a huge jump in the number of $2-million-plus candidates in both parties between 2014 and 2016, for example. But while Democrats’ numbers have held steady or improved from the high levels they had in 2016, Republican numbers have collapsed. The 17 GOP candidates that we project will raise at least $2 million this year is down from 64 in 2016. (All figures are adjusted for inflation.)
Democrats’ fundraising advantage is widespread
Number of candidates on Nov. House ballots raising …
At least $500K by 20 days before the election
At least $1M by 20 days before the election
At least $2M by 20 days before the election
Year
Dem.
GOP
Dem.
GOP
Dem.
GOP
1998
93
128
22
37
3
4
2000
123
153
48
61
4
9
2002
108
151
38
37
8
6
2004
134
178
48
66
8
10
2006
156
172
62
65
13
13
2008
190
168
84
60
18
10
2010
200
220
95
87
20
15
2012
170
227
80
105
18
2
2014
153
197
65
76
23
9
2016
212
249
142
171
80
64
2018*
232
179
144
84
73
17
*2018 totals are extrapolated based on fundraising through Sept. 30.
All figures are adjusted for inflation.
Source: Federal election commission
The result is a fundraising disparity the likes of which we’ve never seen before — at least not in recent years. (Our data on House fundraising goes back to 1998.) In the average House district, the Democratic candidate has raised 64 percent of the money this cycle,2 or almost two-thirds. Likewise, the Democrat has raised an average of 65 percent of the money in districts rated as competitive3 by the Cook Political Report. In all previous years in our database, no party had averaged more than 56 percent of the money in these competitive districts.
Democrats’ fundraising dominance has no recent precedent
Democrats’ share of two-party campaign contributions
All House districts
Competitive House districts
Year
Average
Median
Average
Median
1998
47%
42%
51%
49%
2000
50
49
49
50
2002
49
41
46
44
2004
47
44
49
50
2006
54
54
51
50
2008
60
66
49
49
2010
47
49
54
53
2012
46
43
48
48
2014
48
46
56
55
2016
49
46
52
54
2018*
64
71
65
68
*Based on data available as of Oct. 18, 2018. For other years, totals are based on data available as of Election Day. The data covers individual campaign contributions only, and not candidate self-funding or donations from PACs or outside groups.
Source: Federal Election Commission
The fundraising numbers are so good for Democrats — and so bad for Republicans — that it’s hard to know quite what to make of them. From a modeling standpoint, we’re extrapolating from years in which fundraising was relatively even, or from when one party had a modest edge, into an environment where Democrats suddenly have a 2-1 advantage in fundraising in competitive races. Moreover, this edge comes despite the fact that a large number of these competitive races feature Republican incumbents (incumbents usually have an easier time raising money than challengers) and that most of them are in red terrain.
If Democrats beat their projections on Nov. 6 — say, they win 63 House seats, equalling the number that Republicans won in 2010, an unlikely-but-not-impossible scenario — we may look back on these fundraising numbers as the canary in the coal mine. That data, plus Democrats’ very strong performances in special elections, could look like tangible signs of a Democratic turnout surge that pollsters and pundits perhaps won’t have paid enough attention to. Right now, in fact, the polls are not showing a Democratic turnout advantage. Instead, based on a comparison of likely-voter and registered-voter polls, they’re projecting roughly equal turnout between the parties, with Republicans’ demographic advantages (older, whiter voters typically vote at higher rates in the midterms) counteracting Democrats’ seemingly higher enthusiasm. If turnout among Democratic-leaning groups actually outpaces turnout among Republican-leaning ones, Democrats will beat their polls and our projections.
It’s just as easy to imagine the error running the other way, however. Maybe — precisely because fundraising has become easier and candidates are winning contributions from out-of-state and out-of-district donors — fundraising is no longer as meaningful an indicator of candidates’ grassroots appeal or organizational strength. Maybe the demographics of the Republican coalition have changed such that they’ll no longer raise as much money but will still get plenty of votes. Or maybe the GOP can make up for their lack of individual fundraising with more money from outside groups. If that’s the case, our model could overestimate Democrats’ chances. Although, I should note that while there’s a gap between our Lite forecast, which is based on local and national polls only, and our Classic forecast, which also incorporates fundraising and other “fundamentals” data, it’s not an especially large one. (Lite projects Democrats to pick up 36 seats, on average, as compared to 39 seats in the Classic forecast.)
Either way, we’re in somewhat uncharted territory here. For the most part, the various indicators we use in our House forecast tell a consistent story. The generic congressional ballot, district-by-district polling and the past electoral history of midterm years under unpopular presidents are all consistent with a Democratic edge of somewhere between 6 and 10 percentage points in the House popular vote, and with Democrats being reasonably solid but not overwhelming favorites to win a majority of seats. The fundraising data, on the other hand, points toward a massive Democratic landslide.
As a first approximation, the correct approach with data that looks like an outlier is to average it together with the other indicators — not to throw it out. (More often than you might think, the seeming “outlier” proves to be correct and it’s the other data that was off.) And that’s more or less what our model does. But the fundraising data contributes uncertainty to our forecast in a way that our top line probabilities may not capture well.
October 18, 2018
Politics Podcast: Democrats Are Well Positioned In Key Races For Governor
More: Apple Podcasts |
ESPN App |
RSS
| Embed
Embed Code
Our 2018 governors forecast is live! And in the latest installment of “Model Talk” from the FiveThirtyEight Politics podcast team, editor-in-chief Nate Silver talks about the odds in key races, including Florida, Georgia, Ohio and Wisconsin. He also checks in on the latest fundraising numbers and answers questions from listeners.
You can listen to the episode by clicking the “play” button in the audio player above or by downloading it in iTunes , the ESPN App or your favorite podcast platform. If you are new to podcasts, learn how to listen .
The FiveThirtyEight Politics podcast publishes Monday evenings, with occasional special episodes throughout the week. Help new listeners discover the show by leaving us a rating and review on iTunes . Have a comment, question or suggestion for “good polling vs. bad polling”? Get in touch by email, on Twitter or in the comments.
October 17, 2018
Odds Are, Your Next Governor Will Be A Democrat
In the midterm elections for the House and Senate, Democrats are very likely to win the majority of votes, but they face structural barriers to winning the majority of seats. In the House, for instance, we project that Democrats would need to win the popular vote by somewhere in the neighborhood of 5 to 7 percentage points1 to be favored to actually take control of the chamber, a result of partisan gerrymandering after the 2010 election and Democratic voters’ tendency to cluster in dense, urban districts. And, of course, Congress isn’t the Democrats’ only problem: In the 2016 presidential election, Hillary Clinton won the popular vote by almost 3 million votes but lost the Electoral College.
In gubernatorial races, however, there’s no gerrymandering or Electoral College to worry about. So in some ways, they’ll make for the purest test of whether there really is going to be a “blue wave” this year.
And in FiveThirtyEight’s gubernatorial forecasts, which we (finally!) launched on Wednesday, the gubernatorial news is good for Democrats. They are projected to wind up with governorships in states representing about 60 percent of the U.S. population, compared with 40 percent for Republicans.
True, Democrats will have a hard time winning the majority of states, because the GOP is projected to do well in small states such as Wyoming. But the number of states is a stupid metric: Being the governor of Texas is a hell of a lot more important than being the governor of Vermont. Higher-population states, because they have more seats in the House, will also be more important in the redistricting process that will take place after the 2020 elections, when governors elected this year will still be in office.2
So as our measure of gubernatorial success, we’ll simply be counting up the number of people projected to be under each party’s control. Texas will count 44 times as much as Vermont, because there are 44 times more people there.
Democrats begin with an edge in the population count. Although each party controls seven states from among the 14 governorships that are not up for election this year, the states the Democrats hold are more populous. Specifically, Democrats have about 42 million people under their control — based on the states’ projected populations as of Election Day3 — while the seven Republican states have 26 million people.
In states not on the ballot, Democrats start with a lead
Population in states where governors are not up for re-election
Democratic governor
republican governor
State
Population*
State
Population*
North Carolina
10,365,000
Indiana
6,780,000
New Jersey
9,193,000
Missouri
6,217,000
Virginia
8,590,000
Kentucky
4,532,000
Washington
7,423,000
Utah
3,100,000
Louisiana
4,753,000
Mississippi
3,043,000
Montana
1,059,000
West Virginia
1,859,000
Delaware
971,000
North Dakota
754,000
Total
42,354,000
Total
26,285,000
* Projected as of Nov. 6, 2018.
BASED ON DATA FROM THE U.S. CENSUS BUREAU
Democrats are poised to add to that advantage on Nov. 6, however. Among the nine states with populations of 10 million or more that will elect new governors next month, Democrats are clear favorites in five (California, New York, Illinois, Pennsylvania and Michigan) and modest favorites in a sixth state (Florida). Republicans are clear favorites only in Texas, where incumbent Greg Abbott is likely to be re-elected. The Ohio and Georgia races are toss-ups.
Democrats are well-positioned in high-population gubernatorial races
Race rating according to FiveThirtyEight’s “Classic” forecast as of Oct. 17
State
Population*
Solid D
Likely D
Lean D
Toss-Up
Lean R
Likely R
Solid R
California
39,982,000
✓
Texas
28,320,000
✓
Florida
20,992,000
✓
New York
20,266,000
✓
Illinois
13,106,000
✓
Pennsylvania
13,074,000
✓
Ohio
11,887,000
✓
Georgia
10,516,000
✓
Michigan
10,159,000
✓
Arizona
7,023,000
✓
Massachusetts
6,965,000
✓
Tennessee
6,794,000
✓
Maryland
6,148,000
✓
Wisconsin
5,900,000
✓
Minnesota
5,657,000
✓
Colorado
5,598,000
✓
South Carolina
5,046,000
✓
Alabama
4,960,000
✓
Oregon
4,151,000
✓
Oklahoma
3,976,000
✓
Connecticut
3,677,000
✓
Iowa
3,191,000
✓
Arkansas
3,056,000
✓
Nevada
2,995,000
✓
Kansas
2,968,000
✓
New Mexico
2,136,000
✓
Nebraska
1,943,000
✓
Idaho
1,725,000
✓
Hawaii
1,445,000
✓
N.H.
1,367,000
✓
Maine
1,363,000
✓
Rhode Island
1,080,000
✓
South Dakota
878,000
✓
Alaska
751,000
✓
Vermont
637,000
✓
Wyoming
587,000
✓
* Projected as of Nov. 6, 2018.
BASED ON DATA FROM THE U.S. CENSUS BUREAU
Republicans will make up ground in small and medium-sized states. Indeed, they’re the favorites (specifically, about 2 in 3 favorites, according to our Classic forecast4) to win a majority of states. But Democrats are highly likely — although by no means certain5 — to govern a majority of the population after the election. Even if Republicans win all the toss-up races and all the races our model currently rates as leaning Democratic, they’d come up a little short of a population majority given the states that Democrats have in hand already. Accounting for each state’s population on a probabilistic basis,6 Democrats currently project to have 194 million people under their control after the election, or about 60 percent of the population, compared with 135 million for Republicans.
Methodology-wise, our gubernatorial forecasts are largely the same as our House and Senate forecasts. There are three versions of our model — Lite, Classic and Deluxe — that blend together different forecasting techniques in different ratios. Specifically, these techniques include
A polling average, after polls are adjusted in various ways.
CANTOR, a system that makes projections based on analysis of polling in similar states.
“Fundamentals,” a regression-based method that evaluates non-polling factors such as fundraising that predict election outcomes.
Expert ratings, such as those put together by the Cook Political Report.
In all versions of our forecast, however, polling is by far the largest ingredient in states where there’s a lot of polling. For more detail on our gubernatorial forecasts, including how they differ from our congressional forecasts, see here.
Perhaps the most important difference between gubernatorial races and congressional ones is that partisanship is much less of a factor in governorships. Specifically, it’s only about one-third as important — so, for example, a state that’s typically 15 percentage points more Republican than the country overall in congressional races would only be about 5 points GOP-leaning in gubernatorial races. As a consequence of this, incumbents tend to be favored even when they come from “opposite-colored” states. Republicans Larry Hogan of Maryland and Charlie Baker of Massachusetts are heavy favorites to win re-election despite being in deeply blue states, for example — not just according to the polls, which have them well ahead, but also according to the fundamentals component of our forecast.
On the flip side, few things are truly inevitable in gubernatorial races, especially in states without incumbents. Democrats are competitive in Oklahoma, for example, while Republicans have a fighting chance in Connecticut, despite it being a blue state in a blue year.
How different versions of our governors forecast compare
As of Oct. 17
Inputs
Forecasts
State▲▼
Polls▲▼
CANTOR▲▼
FundAM.▲▼
Experts▲▼
Lite▲▼
Classic▲▼
Deluxe▲▼
Hawaii
D+23.2
D+25.8
D+38.7
D+32.3
D+24.3
D+31.4
D+31.7
New York
D+20.0
D+20.7
D+26.2
D+24.6
D+20.0
D+22.1
D+22.9
Rhode Island
D+11.2
D+22.2
D+33.5
D+7.5
D+11.2
D+19.5
D+15.5
California
D+15.2
D+10.0
D+32.0
D+23.2
D+15.2
D+18.4
D+20.0
Pennsylvania
D+14.6
D+14.4
D+18.4
D+16.4
D+14.6
D+15.4
D+15.8
Illinois
D+17.9
R+4.6
R+9.3
D+12.9
D+17.9
D+12.7
D+12.7
Michigan
D+11.0
D+4.1
D+14.7
D+9.3
D+11.0
D+11.6
D+10.8
Maine
D+7.4
D+5.3
D+18.0
D+2.3
D+7.4
D+11.3
D+8.3
Colorado
D+7.7
D+3.5
D+15.4
D+7.5
D+7.6
D+10.4
D+9.4
Minnesota
D+7.9
D+5.1
D+17.4
D+11.1
D+7.9
D+10.2
D+10.5
Iowa
D+4.9
D+2.0
D+14.7
D+0.8
D+4.9
D+8.0
D+5.6
Oregon
D+4.4
D+18.0
D+14.0
D+5.3
D+4.4
D+7.9
D+7.0
New Mexico
D+6.2
D+5.0
D+10.5
D+7.5
D+6.2
D+7.4
D+7.4
Connecticut
D+5.3
D+5.7
D+8.5
D+5.3
D+5.3
D+6.2
D+5.9
Florida
D+3.7
0.0
D+0.8
D+2.3
D+3.7
D+3.2
D+2.9
Wisconsin
D+5.1
R+8.1
R+7.6
D+0.8
D+5.1
D+2.2
D+1.8
Georgia
R+0.8
R+2.3
D+0.1
R+0.6
R+0.8
R+0.7
R+0.7
Nevada
R+1.0
D+2.3
R+1.4
D+2.3
R+1.0
R+1.1
D+0.1
Ohio
R+0.6
D+1.3
R+4.6
R+0.6
R+0.6
R+1.3
R+1.1
Kansas
D+0.3
R+5.0
R+9.1
R+2.9
D+0.3
R+2.4
R+2.5
Oklahoma
R+3.4
R+10.5
R+12.1
R+5.9
R+3.4
R+6.4
R+6.2
South Dakota
R+0.6
R+7.7
R+16.4
R+3.7
R+4.6
R+10.9
R+8.5
Arizona
R+10.5
R+12.1
R+18.4
R+11.3
R+10.5
R+11.5
R+11.4
Alaska
R+19.6
R+2.3
R+0.6
R+6.9
R+19.6
R+12.3
R+10.5
Vermont
R+12.2
D+0.3
R+15.7
R+22.5
R+5.5
R+13.4
R+16.4
Tennessee
R+16.6
R+7.7
R+6.0
R+14.9
R+16.6
R+14.3
R+14.5
Idaho
R+11.8
R+10.0
R+18.0
R+21.9
R+11.0
R+15.0
R+17.3
N.H.
R+14.0
R+8.0
R+20.8
R+7.7
R+14.0
R+16.1
R+13.3
South Carolina
R+18.9
R+4.0
R+13.0
R+18.2
R+18.9
R+16.1
R+16.8
Alabama
R+19.1
R+7.9
R+16.3
R+21.5
R+15.1
R+17.1
R+18.6
Maryland
R+18.8
R+0.6
R+15.5
R+11.3
R+18.8
R+17.6
R+15.5
Texas
R+19.2
R+14.8
R+17.1
R+21.6
R+19.2
R+18.9
R+19.8
Nebraska
R+26.7
R+22.2
R+23.3
R+26.7
R+22.7
R+22.9
Wyoming
R+23.6
R+25.2
R+25.0
R+23.6
R+25.1
R+25.0
Massachusetts
R+36.5
D+1.5
R+19.8
R+28.7
R+36.5
R+29.3
R+29.1
Arkansas
R+36.5
R+17.7
R+22.7
R+29.7
R+35.6
R+30.4
R+30.1
We’ll cover the most interesting gubernatorial races on an individual basis in subsequent updates, but here are a few comments about races that I know people will have questions about. Democrats Andrew Gillum of Florida and Stacey Abrams of Georgia are striving to become the first African-American governors of their respective states and the first elected anywhere in the South since Douglas Wilder of Virginia in 1989. Gillum has had a small but fairly consistent lead in the polls, and our model gives him a 70 percent chance (about 7 in 10) of winning. Abrams is in a toss-up race that tilts ever-so-slightly toward her Republican opponent, Brian Kemp. The race is close enough that voter registration protocols in Georgia, which Kemp oversees as secretary of state, could make the difference. (Check out my colleague Perry Bacon Jr.’s article on these races for more detail.)
Although there generally isn’t a big conflict between polls and fundamentals in our gubernatorial forecasts, there are a couple of high-profile races where fundamentals nudge the forecast toward the GOP. In Wisconsin, incumbent Republican Scott Walker trails Democrat Tony Evers by 5 points in our polling average, but the fundamentals think he “should” narrowly win re-election. The Classic version of our model evaluates this race as leaning Democratic, but with Walker having a better chance than polls alone would suggest. And in Kansas, where the controversial Republican secretary of state, Kris Kobach, is running, polls show a true dead heat against Democrat Laura Kelly, but the model classifies the race as leaning Republican on the basis of the fundamentals.
2018 Governors Forecast
ELECTION 2018
FiveThirtyEight
House forecast
Senate
Governor
Midterms coverage
More politics
Latest polls
Generic ballot
Trump approval
How do you like your governor forecast?
Lite
Keep it simple, please — give me the best forecast you can based on what state and national polls say
Classic
I’ll take the polls, plus all the “fundamentals”: fundraising, past voting in the state, historical trends and more
Deluxe
Gimme the works — the Classic forecasts plus experts’ ratings
Forecasting the races for governor
Updated Oct. 2, 2018, at 10:18 AM
195 million
Population forecasted to be governed by Democrats
134 million
Population forecasted to be governed by Republicans
23.7
Average states governed by Democrats across our simulations
26.2
Average states governed by Republicans across our simulations[a]
Our forecast for every race for governor
What share of the national population Democrats and Republicans are expected to govern, based on the most likely winner in each of the 36 gubernatorial elections taking place in 2018 and who controls the 14 seats not up for election.
59.0% of thepopulation
WA
VA
NC
NJ
NY
PA
CA
MI
IL
FL
WA
VA
NC
NJ
NY
PA
CA
MI
IL
FL
41.0%
IN
MA
TX
TN
AZ
OH
GA
IN
MA
TX
TN
AZ
OH
GA
500,000 RESIDENTS
No election
Solid D
Likely D
Lean D
Toss-up
Lean R
Likely R
Solid R
No election
≥95%
≥75%
≥60%
≥60%
≥75%
≥95%
AK
AK
CA
CA
CO
CO
ID
ID
IL
IL
IA
IA
KY
KY
LA
LA
MD
MD
MN
MN
MO
MO
NY
NY
OR
OR
TN
TN
TX
TX
VA
VA
WI
WI
AL
AL
AZ
AZ
AR
AR
IN
IN
KS
KS
ME
ME
CT
CT
DE
DE
GA
GA
HI
HI
SC
SC
SD
SD
MA
MA
MI
MI
MS
MS
NE
NE
NV
NV
NH
NH
NJ
NJ
NM
NM
NC
NC
ND
ND
RI
RI
OH
OH
OK
OK
PA
PA
FL
FL
MT
MT
UT
UT
VT
VT
WA
WA
WV
WV
WY
WY
Who’s ahead in each race
The closer the dot is to the center line, the tighter the race. And the wider the bar, the less certain the model is about the outcome.
KEY
AVERAGE
80% CHANCE OUTCOME FALLS IN THIS RANGE
StateCandidatesIncumbentExpected margin of victory
D+75
D+50
D+25
R+25
R+50
R+75
Nevada
–
Sisolakvs.Laxalt
R+0.1
Georgia
–
Abramsvs.Kemp
R+0.6
Ohio
–
Cordrayvs.DeWine
R+1.5
Wisconsin
R
Eversvs.Walker
D+2.2
Kansas
–
Kellyvs.Kobach
R+2.5
Florida
–
Gillumvs.DeSantis
D+3.5
Oklahoma
–
Edmondsonvs.Stitt
R+6.4
Connecticut
–
Lamontvs.Stefanowski
D+7.3
New Mexico
–
Lujan Grishamvs.Pearce
D+7.3
Iowa
R
Hubbellvs.Reynolds
D+7.8
Oregon
D
Brownvs.Buehler
D+9.4
Maine
–
Millsvs.Moody
D+10.4
Colorado
–
Polisvs.Stapleton
D+10.4
South Dakota
–
Suttonvs.Noem
R+10.8
Michigan
–
Whitmervs.Schuette
D+11.5
Arizona
R
Garciavs.Ducey
R+11.8
South Carolina
R
Smithvs.McMaster
R+11.9
Alaska
I
Begichvs.Dunleavy
R+12.0
Minnesota
–
Walzvs.Johnson
D+12.5
Illinois
R
Pritzkervs.Rauner
D+12.6
Vermont
R
Hallquistvs.Scott
R+13.6
Tennessee
–
Deanvs.Lee
R+14.3
Idaho
–
Jordanvs.Little
R+15.0
Pennsylvania
D
Wolfvs.Wagner
D+16.1
New Hampshire
R
Kellyvs.Sununu
R+16.6
Alabama
R
Maddoxvs.Ivey
R+17.1
California
–
Newsomvs.Cox
D+17.7
Maryland
R
Jealousvs.Hogan
R+17.7
Texas
R
Valdezvs.Abbott
R+18.7
Rhode Island
D
Raimondovs.Fung
D+19.4
New York
D
Cuomovs.Molinaro
D+22.2
Nebraska
R
Kristvs.Ricketts
R+22.6
Wyoming
–
Thronevs.Gordon
R+24.8
Massachusetts
R
Gonzalezvs.Baker
R+29.3
Arkansas
R
Hendersonvs.Hutchinson
R+30.3
Hawaii
D
Igevs.Tupola
D+31.4
Our latest coverage
10 HOURS AGO
Young Voters Might Actually Show Up At The Polls This Year
By Geoffrey Skelley
2 DAYS AGO
Why The House And Senate Are Moving In Opposite Directions
By Nate Silver
OCT 12
Can Heitkamp Pull Off A Second Upset In North Dakota?
By Janie Velencia and Dhrumil Mehta
How this forecast works
Nate Silver explains the methodology behind our 2018 midterm forecasts. Read more …
State populations are based on forecasted populations as of Election Day (Nov. 6, 2018), using data extrapolated from the American Community Survey and the Census Bureau’s U.S. population Clock and factoring in recent growth rates in each state.. The third-party and independent candidates listed represent our best approximation of who will appear on each state’s general election ballot. The candidates listed will update as each race is finalized; some listed candidates may not ultimately qualify for the general election.[b][c][d]
Forecast models by Nate Silver. Design and development by Jay Boice, Aaron Bycoffe, Rachael Dottle, Ritchie King, Ella Koeze, Andrei Scheinkman, Gus Wezerek and Julia Wolfe. Research by Andrea Jones-Rooy, Dhrumil Mehta, Mai Nguyen and Nathaniel Rakich. Notice any bugs? Send us an email.
RELATED STORIES
Young Voters Might Actually Show Up At The Polls This Year
Young Voters Might Actually Show Up At The Polls This Year
Why The House And Senate Are Moving In Opposite Directions
Why The House And Senate Are Moving In Opposite Directions
Can Heitkamp Pull Off A Second Upset In North Dakota?
Can Heitkamp Pull Off A Second Upset In North Dakota?
Democrats Have A Chance To Win One Of The Reddest Districts In The Country
Democrats Have A Chance To Win One Of The Reddest Districts In The Country
Politics Podcast: Two Forecasts Diverged In A Wood
Politics Podcast: Two Forecasts Diverged In A Wood
Get more FiveThirtyEight
Newsletter
Videos
Podcasts
Twitter
Facebook
Data
RSS
Contact
Jobs
Masthead
About Nielsen Measurement
Terms of Use
Privacy Policy
Your California Privacy Rights
Children’s Online Privacy Policy
Interest-Based Ads
© 2018 ABC News Internet Ventures. All rights reserved.
Nate Silver's Blog
- Nate Silver's profile
- 724 followers
