2018 Congressional Generic Ballot and House Polls Megathread - the original

[Brittain33]

It is not I who has placed great emphasis on the generic vote this far out.  But the D advantage has been headed down.

Except for 2010 in midterms the GOP has run 3 to 4 points better than the final RCP average. We shall see.

Do you think Dems will underperform because they didn’t send out Moore’s yearbook solicitation of teenagers for independent scientific handwriting analysis, Arkansas Yankee? I remember your predicting that would doom Doug Jones. 😂

[Matty]

[Dr Oz Lost Party!]

That's a very odd way of asking questions...

Can we even count that as a generic ballot poll?

[Hindsight was 2020]

That's a very odd way of asking questions...

Can we even count that as a generic ballot poll?

Yeah what do we do with the 9% that want republicans independent of Trump?

[GeorgiaModerate]

That's a very odd way of asking questions...

Can we even count that as a generic ballot poll?

I don't see why not; I'd count it as D 45, R 43.  It's just more nuanced than the typical GCB question, just like some Trump approval polls separate strong approval/disapproval from somewhat approve/disapprove, but others don't.

[LimoLiberal]

That's a very odd way of asking questions...

Can we even count that as a generic ballot poll?

I don't see why not; I'd count it as D 45, R 43.  It's just more nuanced than the typical GCB question, just like some Trump approval polls separate strong approval/disapproval from somewhat approve/disapprove, but others don't.

Previous CBS poll had D+9. Dems losing ground in both internet and high-quality polls with their 538 lead at its lowest point since May.

[UncleSam]

so basically Dems are at +1 and most Rs support Trump, with lots of undecided voters. To convert that into a 7 point win Dems need to win Indy’s and undecided voters about 3-1 (which is roughly in line with their best performances in special elections). So a flip of the house is possible but unlikely under that scenario.

The Trump support number is very important though. It indicates that the Senate could be every bit as tough a lift for Dems as expected - I did not anticipate supporting Trump to have such a high portion.  Rs who want an independent voice are the best pickup opportunities for Manchin etc, and if they’re only 9% of the population it’s hard to see him or other deep red state Dems winning.

[Dr Oz Lost Party!]

I think that 9% could pose a problem for the GOP. The Dems are running a lot of moderates and progressives in different districts they should try to target the disenfranchised Republicans like the 9% in the poll.

I still expect undecideds to break heavily for the Democrats.

Also, notice how they're using YouGov. They, along with Reuters, have been driving the GCB for the past couple of weeks.

[Bandit3 the Worker]

Ipsos is a joke anymore.

[KingSweden]

That's a very odd way of asking questions...

Can we even count that as a generic ballot poll?

I don't see why not; I'd count it as D 45, R 43.  It's just more nuanced than the typical GCB question, just like some Trump approval polls separate strong approval/disapproval from somewhat approve/disapprove, but others don't.

Previous CBS poll had D+9. Dems losing ground in both internet and high-quality polls with their 538 lead at its lowest point since May.

Looks like this is YouGov rather than SSRS, so while both are affiliated with CBS it is not, technically, the same poll.

[Gustaf]

You can't count that as a GCB poll, because these options are mutually exclusive. Any actual congressional race can pit EITHER a Republican independent from or one loyal to Trump versus EITHER a liberal or a moderate Democrat.

Without knowing the distribution of races or the breakdown of these groups if their preferred option isn't on the ballot you don't know much.

[Hindsight was 2020]

Honestly RCP and 538 don't have a poll that is younger than 5 days ago. There has been a really dearth of polling

[Dr Oz Lost Party!]

I wonder how the polls will react with the Farm Bill Collapse and Santa Fe? Probably not much or at all, but I can hope.

[Matty]

http://polling.reuters.com/#!response/TM1212Y17/type/week/filters/PD1:1/dates/20180101-20180521/collapsed/true

GOP leading according to newest reuters poll if I am reading it right

[Bandit3 the Worker]

http://polling.reuters.com/#!response/TM1212Y17/type/week/filters/PD1:1/dates/20180101-20180521/collapsed/true

GOP leading according to newest reuters poll if I am reading it right

Reuters's new model is already discredited.

[UncleSam]

http://polling.reuters.com/#!response/TM1212Y17/type/week/filters/PD1:1/dates/20180101-20180521/collapsed/true

GOP leading according to newest reuters poll if I am reading it right

Reuters's new model is already discredited.

What do you mean by discredited? As in you don’t like the results or there are known statistical issues with it? Do you have methodological problems with it? I’m not familiar with how Reuters does their poll.

[BudgieForce]

http://polling.reuters.com/#!response/TM1212Y17/type/week/filters/PD1:1/dates/20180101-20180521/collapsed/true

GOP leading according to newest reuters poll if I am reading it right

I just...uhm...what?

[KingSweden]

http://polling.reuters.com/#!response/TM1212Y17/type/week/filters/PD1:1/dates/20180101-20180521/collapsed/true

GOP leading according to newest reuters poll if I am reading it right

Reuters's new model is already discredited.

What do you mean by discredited? As in you don’t like the results or there are known statistical issues with it? Do you have methodological problems with it? I’m not familiar with how Reuters does their poll.

I don’t think it’s discredited, but there’s been discussion over how they changed their model. If they did make tweaks, it’s derived multiple results that are very GOP friendly

[Dr Oz Lost Party!]

Yeah, the Pundits are scrutinizing the hell out of Reuters right now. That’s definitely not right.

[Ebsy]

Yes, obviously the Republicans are going to win the congressional ballot by 6 points in the 2018 midterms. We can all go home I guess!

[Politician]

Yes, obviously the Republicans are going to win the congressional ballot by 6 points in the 2018 midterms. We can all go home I guess!

[UncleSam]

I don’t think anyone is suggesting the poll is accurate. The question is, what is it about the methodology that is different? If it’s a matter of having a rosier turnout model for Rs but also just seeing shifting opinions toward Rs, then it should get thrown in the average. If it’s more experimental, than it shouldn’t be thrown in the average yet until we see how it moves and reacts to shifting opinions.

If the generic ballot is actually down to around even, then an R+5.5 or whatever poll isn’t out of the question. It’s also possible that their turnout model tilts a few (or more than a few) points to the right, making the information valuable in change terms of not in absolute or top line terms.

It’s also possible this is just an outlier like we see all the time. Remember that CNN poll with D+20 or whatever? That looks as out of place next to the average as this does now, just in the other direction.

[KingSweden]

I don’t think anyone is suggesting the poll is accurate. The question is, what is it about the methodology that is different? If it’s a matter of having a rosier turnout model for Rs but also just seeing shifting opinions toward Rs, then it should get thrown in the average. If it’s more experimental, than it shouldn’t be thrown in the average yet until we see how it moves and reacts to shifting opinions.

If the generic ballot is actually down to around even, then an R+5.5 or whatever poll isn’t out of the question. It’s also possible that their turnout model tilts a few (or more than a few) points to the right, making the information valuable in change terms of not in absolute or top line terms.

It’s also possible this is just an outlier like we see all the time. Remember that CNN poll with D+20 or whatever? That looks as out of place next to the average as this does now, just in the other direction.

Yeah it’s worth recalling that CNN poll came out when the averages were more like D+10-11

[UncleSam]

I don’t think anyone is suggesting the poll is accurate. The question is, what is it about the methodology that is different? If it’s a matter of having a rosier turnout model for Rs but also just seeing shifting opinions toward Rs, then it should get thrown in the average. If it’s more experimental, than it shouldn’t be thrown in the average yet until we see how it moves and reacts to shifting opinions.

If the generic ballot is actually down to around even, then an R+5.5 or whatever poll isn’t out of the question. It’s also possible that their turnout model tilts a few (or more than a few) points to the right, making the information valuable in change terms of not in absolute or top line terms.

It’s also possible this is just an outlier like we see all the time. Remember that CNN poll with D+20 or whatever? That looks as out of place next to the average as this does now, just in the other direction.

Yeah it’s worth recalling that CNN poll came out when the averages were more like D+10-11

Sure, but the difference was still 10 points or so off the average. R+5.5 is about the same in the other direction - should’ve been clear I meant adjusted for the difference in average.

[GeorgiaModerate]

FWIW, if you switch the filter on the Reuters poll to 5-day rolling average (which is used by the 538 aggregator), it shows R+1.4 (38.1-36.7).  Still surprising, but less so than the R+5.5 shown the other way.