Before presenting the latest polling, I want to take a look at other data and its implications for the midterms. With much more early voting underway, we now have actual voting behavior to look at (at least in terms of who has been voting, although of course we don’t know how they all voted). However, even before considering that, let’s take a look at something with known outcomes: the post-Dobbs special elections that took place in the context of Roe being officially overturned and GOP-controlled states rushing to pass new abortion restrictions or bans.
Special Elections: Special elections can be an indicator of things to come, because they show how people in different states or districts actually chose to vote post-Dobbs. However, even this sort of data needs some caveats. Events after those elections could have shifted voter intentions, and turnout is typically much lower in such elections than for general elections. This latter point is likely favorable to the Democrats, though, because it’s precisely older voters — skewing more conservative — who are most likely to vote in specials.
And overall, the post-Dobbs specials are very positive for Democrats. The Fivethirtyeight website posted a good summary of Congressional special elections in 2022. Prior to the official release of Dobbs, Republicans were overperforming against the partisan lean of their districts by an average of 2 pts (consistent with the narrative the the GOP should outperform in the midterms with Democratic president). Across 4 elections post-Dobbs, Democratic overperformance was 9% on average, ranging from +6 to +12. That’s a huge swing, and it doesn’t even include the Alaska special election that sent Peltola to the House as the first Democratic representative from Alaska in nearly 50 years. That election is trickier to interpret due to the ranked-choice voting and the infighting between two GOP candidates. However, the surprise Kansas vote to uphold abortion rights, by a stunning majority of +18 pts in the face of predictions of a narrow anti-choice win, indicates that abortion rights are a huge factor in this cycle, unlike any in recent memory. This is almost certainly due to pro-choice voters suddenly becoming as engaged and focused on this issue as anti-choice voters have traditionally been, and Democrats now have a large advantage on this issue. One important caveat is in order, though: all of these elections took place over 2 months ago. This means that the intensity of voter mobilization over abortion may had time to wane, or that the GOP flood of alarmist messaging about inflation and crime may have swayed some pro-choice voters against the Democrats.
Early Voting: Although early voting does not provide any results prior to election day, it does tell us who’s showing up, not just what people are saying they will do. In some states, early voting is reported by voter registration, which in these polarized times is a very strong indicator of how someone likely voted. For other states, demographic characteristics are reported that can indicate if segments of the electorate favoring Democrats (African Americans, women, etc.) are turning out disproportionately. It’s now well-established that early voting skews Democratic, while Election Day turnout skews Republican. So, we want to see if Democrats are meeting or exceeding their relative turnout in the latest election cycles.
Early numbers here are promising. From this thread, we learn that it appears Democrats are exceeding their share of the early vote not just from 2018 (when the Democrats regained control of the House and Senate) but 2020 as well, and are running further ahead of the Republicans than either of those elections. Further, they’re running ahead of their 2018/2020 performance in key swing states: GA, NV, OH, PA, WI. Overall, it adds up to a 2 million-plus vote advantage for Democrats. A deeper dive shows this is happening despite lower turnout from a key Democratic demographic, the youth vote. So, if younger voters catch up to their participation in the last couple of election cycles, the electorate will trend even more Democratic. Another key advantage for Democrats, courtesy of the GNR: women (who skew Democratic) are an even larger share of 2022 early voters than they were in the 2018 midterms, with a 13% gender gap (56-43).
One possible counter-narrative is that the early vote is not meaningful because these are high-enthusiasm voters who would have voted anyway. But this narrative does not allow for increases in voter enthusiasm among Democratic-leaning demographics — which we already have evidence of, in the form of the special election outcomes noted above. In other words, the Democratic advantage in early voting may simply be confirming the surge in voter enthusiasm we saw in the Democratic overperformance in the special elections. I believe there’s nothing in the early voting numbers that should be cause for alarm, and plenty that is encouraging.
This week’s update: As always, below are brief descriptions and the links that can take you right to the sources for the polls and forecasting I summarize. One last point from the Rosenberg twitter thread I linked to that is important: it’s possible to “game” the polling averages. Some polling firms are partisan or known to show a consistent bias in their results (for either the Democrats or the GOP) while others are recognized as neutral. If large numbers of partisan polls are released in a short time frame for a particular race, they’ll skew the average. Rosenberg believes this has happened this month in GA, NV, and PA.
FiveThirtyEight: Senate ratings and House forecast. Provides estimates of election-day vote based on both a polling average and a “deluxe” model that includes historical trends and expert ratings. (538)
Electoral-vote: Provides a simple last-week polling average for Senate races. (E-V)
RealClearPolitics: Another polling average, but broader than Electoral-Vote’s. (RCP)
270towin: Yet another polling average and algorithm. (270)
Cook Political Report: Expert ratings of each race (solid, likely, lean, or tossup). (Cook)
Sabato's Crystal Ball: Also expert ratings (safe, likely, lean, or tossup). (Sabato)
Inside Elections: Yet more expert ratings (solid, likely, lean, tilt, or tossup). (Inside)
And now, on to the numbers and ratings. These data are current as of Saturday 10/29.
Senate Races
|
E-V |
RCP |
270 |
538 (polls) |
538 (deluxe) |
Sabato |
Cook |
Inside |
Az |
D +2 |
D +1.5 |
D +1.6 |
D 51.0-47.0
|
D 50.5-47.5 |
lean D |
tossup |
tilt D |
FL |
R +8 |
R +7.4 |
R +7.0 |
R 52.5-45.5 |
R 53.6-44.4 |
likely R |
likely R |
likely R |
GA |
D +1 |
R +1.6 |
R +1.0 |
D 50.0-48.5 |
D 49.3-49.1 |
tossup |
tossup |
tossup |
IA |
R +8 |
R +7 |
R +7.4 |
R 54.2-45.8 |
R 56.1-43.9 |
likely R |
likely R |
likely R |
NV |
tie |
R +1.2 |
D +2.0 |
R 48.8-48.6 |
R 48.9-48.5 |
tossup |
tossup |
tossup |
NH |
D +3 |
D +3.4 |
D +3.4 |
D 51.3-48,9 |
D 51.4-46.8 |
lean D |
lean D |
tilt D |
NC |
R +4 |
R +4.5 |
R +3.0 |
R 50.2-47.8 |
R 51.2-46.8 |
lean R |
lean R |
tilt R |
OH |
D +2 |
R +2.3 |
R +2.2 |
R 50.9-49.1 |
R 52.3-47.7 |
lean R |
lean R |
lean R |
PA |
R +1 |
D +0.3 |
D +1.2 |
D 49.9-48.0 |
D 49.5-48.4 |
lean D |
tossup |
tossup |
UT |
R +5 |
R +4 |
R +10 |
R 49.3-41.2 |
R 51.9-38.7 |
likely R |
likely R |
likely R |
WI |
R +1 |
R +3.3 |
R +3.3 |
R 51.5-48.5 |
R 52.1-47.9 |
lean R |
tossup |
tilt R |
Senate overview: Our pundits made almost no changes in their senate race ratings, the sole exception being Cook downgrading AZ from lean D to tossup. Numbers on the purely poll-based sites tended to show a more consistent pattern of most races shifting slightly to the benefit of Republicans. AZ, FL, GA, NH, and PA all shifted in this way. It’s hard to believe that Walker’s dismal debate performance in GA, or the latest revelations about him paying girlfriends for abortions, have somehow benefited him, so maybe this is more about conservative voters holding their noses and committing to Walker (because at least he’s not a sinister America-hating Democrat). In PA, it’s possible that on-the-fence voters have shifted towards Oz because the GOP has persuaded them with their scare campaign about Fetterman’s stroke and fitness for office. Or, it’s possible Rosenberg is right and GOP-biased polls are flooding the zone.
Other races have shown no similar shift. In NV, OH, and WI changes in the polling averages measured no more than 2/10 or 3/10 of a point. In NV and OH, whether it’s the Democrats or the Republicans who are ahead depends on which polling site or average you want to believe in.
The big picture seems to be that there is no big picture: control of the Senate is still on a razor’s edge, and the states that seem most likely to flip (NV and PA) would just see a Democratic pickup offset by a Republican one. Overall Senate forecasts reflect this. 538 sees 48 D and 49 R seats with 3 tossups, and 50 seats for each party the most likely outcome. 270 sees 49-49 with 2 tossups; based on polling averages, RCP shows 49 D and 51 R seats; and E-V sees 50-49 in favor of the Democrats, with one tie.
House overview: Some movement this week in favor of the Republicans. 4 of 6 sites now see the GOP winning outright control.
RCP: R 225, D 173, T 37
270: R 223, D 201, T 11
538: R 219, D 205, T 11
Sabato: R 218, D 195, T 22
Inside Elections: R 212, D 205, T 18
Cook: R 211, D 191, T 33
Gubernatorial races
|
RCP |
270 |
538 (polls) |
538 (deluxe) |
sabato |
cook |
inside |
AZ |
R +3.2 |
R +4.2 |
R 50.8-49.2 |
R 50.8-49.2 |
tossup |
tossup |
tossup |
FL |
R +11.2 |
R +11.4 |
R 53.7-44.2 |
R 54.6-43.4 |
likely R |
likely R |
likely R |
GA |
R +7.8 |
R +7.8 |
R 52.6-46.0 |
R 52.9-45.7 |
lean R |
lean R |
tilt R |
KS |
D +2.0 |
D +2.0 |
D 50.5-46.2 |
D 49.3-47.5 |
tossup |
tossup |
tossup |
NV |
R +2.4 |
R +1.0 |
R 49.4-48.0 |
D 49.0-48.4 |
tossup |
tossup |
tossup |
OK |
D +2.5 |
D +4.0 |
R 49.5-46.8 |
R 53.3-42.9 |
likely R |
likely R |
likely R
|
OR |
R +1.7 |
R +0.6 |
D 41.6-41.3 |
R 41.1-41.1 |
tossup |
tossup |
tossup |
PA |
D +6.7 |
D +12.8 |
D 53.5-44.5 |
D 53.5-44.6 |
likely D |
likely D |
lean D |
TX |
R +8.3 |
R +8.4 |
R 53.2-44.7 |
R 54.9-42.9 |
likely R |
likely R |
solid R |
WI |
D +0.5 |
D +0.5 |
D 49.4-49.1 |
D 49.6-48.9 |
tossup |
tossup |
tossup |
Gubernatorial Overview: Shifts in the governors’ races seem to be more of a mixed bag than in the Senate races. There were no changes in pundit ratings. In polling, however, shifts in AZ, FL, and GA favored Republicans. However, WI showed a slight swing back towards the Democrats, while other states showed no meaningful change or else mixed outcomes. A new race to pay close attention to is in Oklahoma — yes, what we think of as ruby-red Oklahoma — where Democratic candidate Joy Hofmeister is leading GOP incumbent Kevin Stitt in the RCP and 270 polling averages. Hofmeister was in fact the GOP state superintendent of education and switched parties specifically to beat Stitt, which probably explains most of her competitiveness. It also didn’t hurt that she received a rare unanimous endorsement from leaders of the five main Native American peoples in Oklahoma, who make up nearly 10% of the state population.
The governor’s races also provide a good example of why you should never, ever freak out about one poll’s results — and how a single poll can also distort the overall average. Outlier values get a lot of attention because they’re surprising, but the very fact of being an outlier suggests that they’re likely wrong. Our case in point here is the AZ governor’s race. Consider the 270 average, which shows Lake beating Hobbs by an avg of 4.2 pts. This is actually a notable jump from last week, when Lake’s lead was only 2.4. What happened? Checking the underlying numbers at the website, the current average is based on 5 polls from within the last week, which had leads of Hobbs +1 and Lake +3, +4, +4 — and +11. The site’s algorithm dutifully averaged all 5 polls. But the +11 result is radically different from the others, and therefore probably wrong. Discarding the outlier, we get an average lead of +2.5 for Lake, which is almost identical to last week’s +2.4. In contrast, DeSantis’ lead in FL is also based on 5 recent polls, in which DeSantis leads (unfortunately) by 10, 10, 11, 12, and 14 pts. So DeSantis’ average is not being skewed by any one outlier. The moral: never make too much of a single poll result for any politician (or for anything else, for that matter).
Next week: our final look before the election! I’ll update through Sunday numbers to capture as much last-minute information as possible.