This morning, the FiveThirtyEight polling average has Biden up by +8.6 points (51.1-42.6). After narrowing in late July and early August from a high of Biden +9.6 on July 2nd, Biden is slowly but steadily widening his lead again. With the Democratic National Convention wrapping up last night, you can be forgiven for thinking this might be a convention bounce for the former vice president. But, all of the polls released thus far (including the one released so far this morning) were in the field days ago. None were in the field past August 18th, which means a few could have been impacted by the first couple days of the convention but that impact on those few polls is not likely to be significant. Next week’s polls should tell us about any convention bounce for Biden, although considering the political climate and the polling data that consistently shows almost all voters already strongly attached to one candidate, I doubt we’ll see the kind of convention bounce (for either candidate) that we often saw in the past. What explains Biden’s increase in support lately is more likely related to the A-rated pollsters releasing new polls over the past ten days while lower-rated (and some pretty awful) pollsters had July almost all to themselves.
Let’s take a look at the polls for the past week.
On the national presidential polling front, two A-rated pollsters released results of surveys this week on the presidential race. ABC News/Washington Post (A+) has Biden up by 10 points among likely voters (54-44) and by 12 among registered voters (53-41). NBC News/Wall Street Journal (A-) has Biden up by nine points (50-41). And CNN/SSRS popped up again showing a much tighter race and making spurious claims about battleground states (see my May 22nd post, which I am going to republish later today for reference). SSRS (B/C) has Biden up by just four points (50-46). I am not going to report the battleground states claims because they are misleading (again, see the May 22 post). However, that does not mean that the national sample is incorrect, but considering the polling averages it does appear to be an outlier. I am not sure what SSRS is doing to have consistently lower margins on the national race than both the A-rated pollsters and the FiveThirtyEight polling averages, and it’s not due to likely voter screens (since this is a poll or registered voters) or partisan breakdown among registered voters nationally (which they get almost exactly correct for this week according to Gallup). My guess is that this is related to how they complied their call list. They don’t mention it in their methodology, but my guess is that either (1) they are not using public voter files to create their call list or (2) there is something wrong with their weighting protocols, particularly where they oversampled the 15 battleground states (or both).
Here are this week’s other national presidential polls for your review:
Data for Progress (B-): Biden 50%, Trump 41%; Biden +14
Morning Consult (B/C): Biden 49%, Trump 39%; Biden +10
YouGov (B): Biden 50%, Trump 40%; Biden +10
Ipsos (B-): Biden 48%, Trump 36%; Biden +9
CNN/SSRS (B/C): Biden 50%, Trump 46%; Biden +4
Echelon Insights (N/A): Biden 51%, Trump 38%; Biden +13
Rasmussen/Pulse Opinion Research (C+): Biden 48%, Trump 44%; Biden +4
HarrisX (C): Biden 45%, Trump 39%; Biden +6
Redfield & Wilton Strategies (N/A): Biden 48%, Trump 41%; Biden +7
There was not a lot of good state-level polling this week. There was a shockingly terrible poll by OnMessage, Inc., a polling firm run by former Louisiana Gov. Bobby Jindal’s (R) former political team and – for this poll – working for Heritage Action for America a nonprofit sister organization to the Heritage Foundation. The results were, in my opinion, clearly massaged for the desired results. This poll is similar to the 15-state battleground state sub-sample problem with last month’s CNN poll referenced above. The sample is 400, but this covers four different states. The poll is a mess, perhaps intentionally so.
There was one A+ pollster with a state-level poll this week and it is something to keep an eye on. Muhlenberg College surveyed Pennsylvania and found Biden leading by four points (49-45). The four point margin is interesting because it is the same margin and nearly the same percentage vote as OnMessage’s (50-46). It is also the third poll in a row showing Biden +4, while the FiveThirtyEight polling average is Biden +6.1. The 6.1 average, which includes the three polls referenced above, is down from Biden +8.1 on July 1st. The race seems to be narrowing in the Keystone State. We need to keep an eye on this state.
Senate Polling This Week
There were a number of single polls in half a dozen states this week. Remember, single polls can be outliers or simply wrong (even at 95% confidence, which most pollsters use, this still means that one out of every 20 polls will fall outside of the MOE). However, we got two polls in the Georgia regular Senate election this week. Garin-Hart-Yang Research Group (B/C) has Democrat Jon Ossoff leading GOP Sen. David Perdue by two points (48-46). Public Policy Polling (B) has it even (44-44). This race remains close and a very possible Democratic pick-up in November.
Here are the rest of this week’s Senate polls:
Iowa: Public Policy Polling (B): Greenfield (D) 48%, Ernst (R) 45%; Greenfield (D) +3
Maine: Public Policy Polling (B): Gideon (D) 49%, Collins (R) 44%; Gideon (D) +5
Michigan: Tarrance Group (B/C): Peters (D) 49%, James 44%; Peters (D) +5
Arizona: OnMessage Inc. (B/C): Kelly (D) 48%, McSally (R) 48%; EVEN
Texas: YouGov (B): Hegar (D) 37%, Cornyn 44%; Cornyn (R) +7
N.C.: E. Carolina Univ. (B/C): Cunningham (D) 44%, Tillis (R) 40%; Cunningham (D) +5
Note on House Seats
There is still not a lot of Congressional District polling – at least not much that is readily accessible. But, several analysts and forecasters are following the House contests. Inside Politics sees 64 competitive seats this year, including 30 that are either Toss-Ups or Tilts (IP uses the “Tilt” category between “Lean” and “Toss-Up”). Their outlook on the race has recently changed: “Our previous projected House range was a Republican gain of five seats to a Democratic gain of five seats. We are adjusting that in favor of Democrats to a most likely range of no net change to a Democratic gain of 12 seats. And that might still be understating Democratic gains.”
This is consistent with what Cook Political Report expects, although they are a bit more conservative. CPR now believes that the Democrats will pick up 10 seats in the House. CPR does not use the “Tilt” category, but sees 28 seats as Toss-Ups. Sabato’s Crystal Ball and Decision Desk are a bit more conservative, seeing 14 seats as Toss-Ups, although they have a lot of seats in the “Lean” categories. The data and viz at Decision Desk is worth checking out if you are interested; they have forecasts and maps for president, Senate, and House.
Check out the Texas Trifecta Brief, scheduled for release next week. CPR has 12 Congressional Districts that are competitive (meaning anywhere from Toss-Up to Likely D/R). Only nine states have more than 12 Congressional seats. Texas could be really interesting this year.
Don’t Feed Your Anxiety
That’s it for this week. Have a great and relaxing weekend! Next week is the RNC convention, but don’t feel tempted to watch it. It will be a Trumpian ego and hate fest, so if that will get your blood pressure up or increase your anxiety over the state of our Republic and the upcoming election do NOT watch it. It’s not important, and you can always hear about what happened later in the news, from your friends, on social media (not necessarily a place to hang if you want to avoid stress), or here (although, in general, I am only going to discuss the conventions as far as they impact the polling). Practice self-care and pace yourself; we need your help this fall!