DDHQ early forecast for 2018 House: D's gain 12 seats

[Oryxslayer]

I guess Clinton-district Republicans like Rohrabacher can sleep tight and continue taking pics with holocaust deniers.

This is what got me confused and sent me diving into the 'juice' of the model. They presently project CA-39 to flip, and CA-48 to not which goes against pretty much all other projection sites who currently put the 48th at Tossup while the 39th at lean R.

The problem, appears to be that the model might be weighting past results by more then everyone else here - aka their model is incumbent friendly. Whether this is a bad thing or a good thing is your opinion.

Examining the previous example of the California seats showed me why it seems favored towards incumbent results. There is a PVI difference of 4% R between CA - 39 and CA - 48.  However, it isn't going by a pure PVI scale. If it was, then NY - 24 and WA - 08 would be blue. Rather it seems to be based off of Incumbent results last year.

Take AZ - 02, a seat we know to be competitive since the democrat constantly wins the seat but by less then their national margin. Before 2016, it was Tossup, after 2016, it was R+1. In every election since 2010, the candidate who won by less then 3% margin - even in the 214 wave. Except in 2016, the Democrats had no hope of retaking the house so they didn't invest in taking down McSallys warchest, so the Republicans won 57-43. They currently give the dem 50% in their forecast with a 47% chance of victory on a 8% national lead, when this is the kind of district Dems should be picking up easily. McSally even released rumors a while ago that she was below "generic D" in her district. So it seams, they are weighting past results a little to much, when many seats from past waves that have flipped were ones not given much thought in previous years.

I think Sabato explains best why using past result might not be the best predictor in many cases here:

http://www.centerforpolitics.org/crystalball/articles/for-house-republicans-past-performance-is-no-guarantee-of-future-results/

[Pericles]

TX-23 will not flip, Hurd is too good a fit imo.

He barely won in 2016 he'll lose in 2018.

[McGovernForPrez]

DDHQ is a joke, we've known this since the Michigan and Missouri mishaps in March '16. That being said, the Republicans will keep the house, even if only by a seat or two. GA-6 Special showed us that dems lack the skill needed to gain 24 seats.

Yeah they're hacks. I don't think Dems are guaranteed to take back the house with a +8 margin but they sure as hell are getting more than 12. Any model predicting that small a gain must be absolute garbage.

[Yank2133]

Yeah, this is crap.

Democrats may come up short, but they are  going to win more then 12 seats.

[Roronoa D. Law]

So Democrats have a better chance at winning GA-06 than holding MN-01?

[Obama-Biden Democrat]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

[Miles]

Great analysis from DDHQ, as usual!

[Pericles]

For the record, if the Democrats had won the House popular vote in 2016 by 8.5%, they would have gained 14 more seats compared to their actual performance.* However they'd still have fallen short, with a 227-208 GOP majority.

*The seats that'd flip would have been AL-02(which probably won't flip in 2018 as Roby only did so badly by a fluke), CA-10, CA-25, CA-49, CO-06, IA-01, ME-02, MN-02, NE-02, NY-19, NY-22, PA-08, TX-23, VA-10.

[smoltchanov]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

48 seats? Come on. Geography, gerrymandering, and hyper-polarization are all working strongly against the Democrats in a way it didn't against the GOP in 2010.

+1. So far i can't get Democrats winning more then 18-20 seats under almost any reasonable assumptions. Too many 75+% Democratic districts, which are absolutely unneccessary for victory, and even harmful, as they make neighbouring districts more Republican-leaning

[smoltchanov]

Great analysis from DDHQ, as usual!

+1.

[Gustaf]

Um, if Democrats actually win the House PV by 8% it obviously won't be a uniform swing.

1.Unopposed incumbents in ultra-safe districts by definition face no swing at all.
2. There are on average more swing voters in swing districts, by definition, so swings are larger there.
3. Most importantly, any swing is a result of the parties' actual efforts in the districts they are targeting so much more likely to occur disproportionately in those places.

[GeorgiaModerate]

If uniform swings were a real thing, Hillary Clinton would be President today.

[windjammer]

This would be pretty pathetic for Dems.

Yes, this is only marginally better in terms of seats than what Democrats got when they won the House PV by ~1% in 2012. I'm not sure if DDHQ wrote on any projected House PV win margin in this article, but the idea that it would be just 2% or so (which is what a +12 seat gain might suggest) seems laughable right now. Democrats have consistently been up 8 - 10 points on the generic ballot for months.

I get that it's just a projection, but it seems pretty weak given all the other evidence we are seeing of a possible wave next year.

Yes, it is like if winning the House by 1 or by 8 makes no difference in this model. That is ridiculous lol.

And by the way, it is important to notice as well that new maps had been adopted between 2012 and 2016. With the new maps in VA and FL, democrats would have netted 2 other seats.
So the 2012 results would have been 203-232.

So this DDHQ is predicting that from a +1 to 8 margin, democrats would only net additional 3 seats? This is ridiculous.

[PragmaticPopulist]

While this is an early prediction, I still see a few questionable projections here. They give Ds a 61% chance of winning Valadao's CA-21 while giving Ds only a 40% chance of winning Rohrabacher's CA-48. IMO Rohrabacher is more vulnerable than Valadao. Valadao is surprisingly resilient for being in a district that leans so Democratic at the presidential level. I'll admit though that his vote for the AHCA may sink him. On the other hand, Rohrabacher acts like he's in a safe R district, continuing to defend Putin and meeting with Julian Assange.

[The Other Castro]

Maybe they'll be wrong, maybe they won't. Midterms are still very far out, and no model right now should be discounted simply because the outcome isn't preferable. That kind of thinking is what gets people into trouble. DDHQ does some good work and shouldn't be dismissed.

[Obama-Biden Democrat]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

48 seats? Come on. Geography, gerrymandering, and hyper-polarization are all working strongly against the Democrats in a way it didn't against the GOP in 2010.

Have you seen the swing in the 2017 congressional races so far? It's something like a 12-15 point average swing from the 2016 congressional results. In SC-05, the Democrat lost by 20 points in 2016 and in 2017, the Democrat only lost 3 points, a 17 point swing. I admit it is early and Republicans may not be as demoralized in 2018. But it much more likely that things get worse for Trump's approval ratings and the GOP congress. Presidential approval ratings are almost always best in their first year, and Trump is still technically in the 6-12 month honeymoon period, and he has just disastrous approval ratings.

The other big problem is Trump is pissed at the GOP congress and has been bashing McConnell on Twitter and has been endorsing insurgent Trumpists in the primaries. This could depress GOP turnout even more as angry Trumpists sit it out to spite the 'establishment'. The other problem is that Trump like Obama in 08, brought out lots of first time unlikely voters. They largely sat out 2010 because Obama was not on the ballot. The same looks likely in 2018.

[Person Man]

Where were they win 2006 or 2010 at this point? I remember everyone in 2005 saying that Democrats would win maybe 6 to 10 seats...

[Since I'm the mad scientist proclaimed by myself]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

48 seats? Come on. Geography, gerrymandering, and hyper-polarization are all working strongly against the Democrats in a way it didn't against the GOP in 2010.

+1. So far i can't get Democrats winning more then 18-20 seats under almost any reasonable assumptions. Too many 75+% Democratic districts, which are absolutely unneccessary for victory, and even harmful, as they make neighbouring districts more Republican-leaning

Have either of you actually looked at the map?

[Since I'm the mad scientist proclaimed by myself]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

48 seats? Come on. Geography, gerrymandering, and hyper-polarization are all working strongly against the Democrats in a way it didn't against the GOP in 2010.

+1. So far i can't get Democrats winning more then 18-20 seats under almost any reasonable assumptions. Too many 75+% Democratic districts, which are absolutely unneccessary for victory, and even harmful, as they make neighbouring districts more Republican-leaning

Have either of you actually looked at the map?

Have you contemplated the PV margin Democrats would need to overcome every institutional disadvantage they have to net 48 f**king seats?

It takes 25-26 seats to take back congress. OP is talking about 48 seats, nearly double that number for Christ's sakes.

Again, have you looked at the map? They'd have to do quite well to get 48, but you don't need any ridiculous seats.

[DrScholl]

Midterms usually defy conventional constraints in more than one place, because most are impacted by enthusiasm and turnout.

[mcmikk]

No one thought Democrats would win all those upstate NY districts or Pennsylvania and Ohio districts in 2006 early on either.

Again, have you looked at the map? They'd have to do quite well to get 48, but you don't need any ridiculous seats.

Man...by March of next year you two will have shifted the conversation to whether or not the Dems will net gain 75 seats and then to 100 seats a few months later.

By October 2018 the only question left will be if the Democratic permamajority will reign for 1,000 years or 10,000 years. Oh and if Queen Harris (peace be upon her) will win by 30 points in 2020 or 70.


Sure anything can happen in politics but the Democrats right now are the Chicago Cubs of polticial Parties. They did win the World Cup last year so sure, anything is possible. Jerk off to your wildest fantasies.

My guesses are about 5700 years and 68 points.

[PragmaticPopulist]

No one thought Democrats would win all those upstate NY districts or Pennsylvania and Ohio districts in 2006 early on either.

Again, have you looked at the map? They'd have to do quite well to get 48, but you don't need any ridiculous seats.

Man...by March of next year you two will have shifted the conversation to whether or not the Dems will net gain 75 seats and then to 100 seats a few months later.

By October 2018 the only question left will be if the Democratic permamajority will reign for 1,000 years or 10,000 years. Oh and if Queen Harris (peace be upon her) will win by 30 points in 2020 or 70.


Sure anything can happen in politics but the Democrats right now are the Chicago Cubs of polticial Parties. They did win the World Series* last year so sure, anything is possible. Jerk off to your wildest fantasies.

I've gone over the house map hundreds of times since the election, and I think the best Democrats can hope for right now is a net gain of 32 or 33 seats, but the potential pickups will likely increase. Yes, gerrymandering and incumbency works against them, but the congressional districts were drawn before it was known that a Republican could win districts like PA-17 and Democrats could win districts like CA-48. While some of these results might be one-offs, I can see a number of districts that were already trending D before 2016 continue their trend, while more recent GOP-trending districts like IL-12 and NY-21 snap back to their pre-2016 behaviors.

[smoltchanov]

That is way too conservative of an estimate, I expect 2018 to be a 2010 like landslide. You could triple or even quadruple that absurdly low number.

48 seats? Come on. Geography, gerrymandering, and hyper-polarization are all working strongly against the Democrats in a way it didn't against the GOP in 2010.

+1. So far i can't get Democrats winning more then 18-20 seats under almost any reasonable assumptions. Too many 75+% Democratic districts, which are absolutely unneccessary for victory, and even harmful, as they make neighbouring districts more Republican-leaning

Have either of you actually looked at the map?

Many times. Probably - more times then you.

[Since I'm the mad scientist proclaimed by myself]

No one thought Democrats would win all those upstate NY districts or Pennsylvania and Ohio districts in 2006 early on either.

Again, have you looked at the map? They'd have to do quite well to get 48, but you don't need any ridiculous seats.

Man...by March of next year you two will have shifted the conversation to whether or not the Dems will net gain 75 seats and then to 100 seats a few months later.

By October 2018 the only question left will be if the Democratic permamajority will reign for 1,000 years or 10,000 years. Oh and if Queen Harris (peace be upon her) will win by 30 points in 2020 or 70.


Sure anything can happen in politics but the Democrats right now are the Chicago Cubs of polticial Parties. They did win the World Series* last year so sure, anything is possible. Jerk off to your wildest fantasies.

CA-10
CA-21
CA-25
CA-48
CA-49
5

AZ-02
UT-03
WA-08
CO-03
CO-06
10

NE-02
KS-03
TX-07
TX-23
TX-32
15

MN-02
MN-03
IA-01
IA-03
MO-2
20

AR-02
MT-AL
FL-15
FL-16
FL-18
25

FL-26
FL-27
GA-06
NC-02
NC-09
30

NC-13
VA-02
VA-10
WV-03
IL-12
35

IL-13
WI-07
MI-07
MI-08
PA-06
40

PA-07
PA-08
PA-16
NJ-07
NY-19
45

NY-22
NY-24
ME-02
OH-01
OH-14
50

Fifty winnable seats. And I left several on the table.

Also: https://uselectionatlas.org/FORUM/index.php?topic=264008.0

[Since I'm the mad scientist proclaimed by myself]

Mega-snip

I know you believe that Dems can't do very well because muh polarization, etc, but come on. This is dumb.