Welcome to week 8 of my tracking of polls and pundits — so you don’t have to do it all yourself! We are now just over 2 weeks from election day. Before we start, I’d like once again to remind everyone of some useful rules of poll-reading:
(1) Never, ever freak out over one poll (or, for that matter, feel elation).
(2) There is always a margin of error. It is literally not possible to eliminate this.
(3) There is always the possibility of systemic error, and we won’t actually know how much and what the main causes are until after the election.
And, based on last week’s discussion of bad-faith “red wave” pollsters working to skew the polling aggregates,
(4) Don’t assume all pollsters are motivated to provide their best-faith estimate of where races stand.
Which brings me to some new points I’d like to make for this week: polls are about intent (respondents answering questions about who they’re going to vote for, and how enthusiastic they are) and likelihood (pollster modeling of likely voters, based on demographics, past election data, etc.). Votes are real behavior. So to really understand what’s going on, we need to look at the most recent voting behavior. There are two potential sources of this: special elections and early voting in this election.
In the case of special elections, Democrats have been overperforming not just against their poll numbers for these elections, but against their vote results in these districts in the 2020 election. Which, you’ll recall, Biden won, sweeping 6 of our current swing states (WI, MI, PA, GA, AZ, and NV, with only NC going to Trump). The Downballot provides an excellent tracker of these results. Across 24 races this year, distributed across the U.S. in large and small red and blue states, Democrats have outperformed their 2020 numbers by 0.8%. Doesn’t sound like a lot? Keep in mind that Biden only won Arizona by 0.3% of the vote, Georgia by 0.2%, Wisconsin by 0.6%. A 0.8% increase more than doubles any of these margins. “But Psychbob!” you say. “Special elections only reflect unique local circumstances and are not predictive of the general election!” And I would agree — if we were just talking about one specific race, or the most recent race. But across a larger number of widely dispersed elections over time, with voting becoming more rigidly polarized in recent election cycles, one starts to see a pattern. One sign of caution, though, is that the 2024 overperformance is much less than the 2023 Democratic overperformance, in which Democrats ran a whopping 5.2% ahead of their 2020 results. So, this indicator seems uncertain.
So what about voting in the current election? Even better! Early voting is underway across the U.S., either by mail or in person or both. However, we don’t know who these people voted for, just that they did vote. Still, it’s possible to track early voting by party registration from states that report it, or to model party affiliation based on other demographics. Targetsmart's early voting dashboard, as of today, estimates a 52.0-39.7 Democratic advantage, or 12.3 points. In the 2020 election, the Democratic advantage was 48.2-41.1, or 7.1 points. So voting to date does seem to reflect all the indicators of greater Democratic enthusiasm and motivation to vote that have already shown up — such as the massive uptick in polls of enthusiasm in Democratic voters when Harris moved to the top of the ticket. Will this hold up to the end of early voting? We don’t know. But as Simon Rosenberg keeps saying on Hopium, I’d rather be us than them.
Overall: A couple of sites show movement of EVs to Harris, but others show continued slight movement towards. Harris’ overall EV count was 257-280 last week across the sites that don’t stick to lots of “toss-ups”, but is now in the 251-276 range. Harris still maintains her 226-219 edge in sites including toss-up ratings, with the exception of RCP. I’m dropping tracking of FL in the state table below; there just hasn’t been any persuasive movement towards Democrats there.
Electoral-vote.com (E-V): Harris 251, Trump 256, ties 31 (MI and NC). Shift towards Trump, with the ties in MI and NC, but WI moves from a tie back into Harris’ column. This site continues to shift back and forth rapidly, with the various swing states moving from one candidate to the other or into ties.
538: Harris 266, Trump 272 (slight shift toward Trump). Harris still has a predicted edge in WI, MI, and NV; Trump in NC, GA, and AZ; and PA is rated even.
The Economist: Harris 273, Trump 265 (movement towards Harris). Harris moves back to a tiny edge in WI and PA.
Princeton Election Consortium: Median of 263 electoral votes for Harris (movement towards Harris). “Moneyball” states are now NC, NV, and WI, with Alaska disappearing from this group as mysteriously as it appeared.
JHK Forecasts: Harris 276, Trump 262 (slight shift again towards Trump). Closest D states are PA, MI and WI and closest R states: NC, GA, AZ. Exactly the same as last week, the week before, the week before...
270towin.com: No change in the “consensus” map: Harris 226, Trump 219, tossups 93.
Sabato’s Crystal Ball: No update — still Harris 226, Trump 219, 93 tossups.
Cook Political Report: No update — still Harris 226, Trump 219, 93 toss-ups.
InsideElections.com: No update — still Harris 226, Trump 219, 93 toss-ups.
CNN: No change — Harris 226, Trump 219, toss-ups 93.
Real Clear Politics (RCP): Same as last week — Harris 215, Trump 219, 104 tossups (MN, WI, MI, PA, NC, GA, AZ, and NV). Their “no tossups” map now has Harris 236, Trump 312, with Trump winning all toss-ups except MN.
State averages/forecasts:
| State | E-V | 538 | 270 | ECON | JHK | RCP | |
|---|---|---|---|---|---|---|---|
| WI | R +1 | Tie | R +0.7 | D +0.3 | R +0.2 | R +0.2 | |
| MI | Tie | D +0.4 | Tie | D +1 | D +0.7 | R +1.2 | |
| PA | D +1 | R +0.2 | D +0.3 | D +0.3 | Tie | R +0.7 | |
| NC | Tie | R +1.2 | R +0.5 | R +1 | R +0.9 | R +0.5 | |
| GA | R +2 | R +2.0 | R +1.8 | R +2 | R +1.8 | R +1.8 | |
| AZ | R +2 | R +2.1 | R +1.7 | R +1 | R +1.8 | R +1.6 | |
| NV | D +2 | D +0.2 | Tie | D +1 | D +0.9 | R +0.8 |
The Senate:
Every site currently has the Rs at 50 to 52 seats. Ds are at 46-50 (except on RCP) with OH as the most consistent toss-up. I’m dropping the tracking of Florida, due to the lack of any substantive movement in favor of the Democrats.
There remains the curious paradox of substantial leads for Democratic Senate candidates in states where the polling/forecast lead for Harris is much narrower. I invite informed speculation on this (or even your random guesses).
Electoral-vote.com: Still calling for D 50, R 50.
The Economist: No change — D 49, R 51.
PEC: No change — D 49, R 51.
JHK: No change — D 49, R 51.
270: Now calling for D 47, R 52 (with NV now predicted as an R pickup), with 1 tossup (OH).
Sabato: No update — D 48, R 51, 1 tossup (OH).
Cook: Still D 46, R 51 (pickups in MT and WV), with 3 tossups (WI, MI and OH).
Inside Elections: Updated 10/18, but still D 48, R 51 (pickups in MT and WV), 1 tossup (OH).
RCP: Now D 44, R 51 (pickups in MT and WV), and 5 tossups (WI, MI, OH, PA, and NV), with the no toss-up map still D 49, R 51.
Senate Averages/Forecasts:
| State | E-V | Econ | 270 | JHK | PEC | RCP | Sabato | Cook |
|---|---|---|---|---|---|---|---|---|
| MT | R +8 | R +4 | R +7.5 | R +4.9 | R +8 | R +7.0 | Lean R | Lean R |
| WI | D +4 | D +5 | D +3.4 | D +4.0 | D +3.5 | D +2.5 | Lean D | Tossup |
| MI | D +4 | D +4 | D +4.2 | D +4.2 | D +5 | D +2.2 | Lean D | Tossup |
| OH | D +1 | D +1 | D +2 | D +0.4 | D +1 | D +0.6 | Tossup | Tossup |
| PA | D +5 | D +5 | D +6.2 | D +5.4 | D +7 | D +2.1 | Lean D | Lean D |
| AZ | D +9 | D +8 | D +7.3 | D +8.3 | D +9 | D +6.3 | Lean D | Lean D |
| NV | D +9 | D +9 | D +9.8 | D +10.3 | D +5 | D +5.4 | Lean D | Lean D |
| NE | I +6 | R +4 | R +1 | R +3.9 | I +3 | NA | Likely R | Likely R |
| TX | R +4 | R +5 | R +4.2 | R +4.2 | R +5.6 | R +4.5 | Likely R | Likely R |
The House:
This week, there’s little difference from last. Some predict Ds, some predict Rs, most have too many tossups and are not committed.
538: 538 projects 217 D, 218 R. Adding up solid/likely/lean seats on both sides results in 212 R, 212 D with 11 tossups.
The Economist: Still predicting 220 D, 215 R.
PEC: PEC gives a D +1.0 edge for control, aligned generic polling.
270toWin: Now at 204 D, 206 R, with 25 tossups (unlike 538, they do not provide a no toss-up forecast).
Sabato: No update; D 206, R 211, with 18 tossups (10 D-held and 8 R-held seats).
Cook: Updated 10/18. D 202, R 206, with 27 tossups (12 D-held and 15 R-held).
RCP: No change — D 196, R 207, with 32 tossups (19 D-held, 13 R-held).
That’s it for this week — check in with me again next Sunday!