Dem
49
GOP
51
We are hoping to get at least five more of these written before the election, because we have a lot to say. That said, they do take a while to put together, and that's on top of the other content, not to mention other responsibilities.
Anyhow, we certainly appreciate folks who read what we write, and agree with us. However, our single-greatest reward, since we are educators first and foremost, is folks who read what we write, and it inspires them to think about some question or issue we raise. Last week, we had an item on how Donald Trump is largely running way ahead of his fellow swing-state U.S. Senate candidates, while Kamala Harris is largely running a bit behind hers. We noted that the gaps we're seeing would be considerably larger than the ones for the first two Trump elections, which could indicate some wonkiness. We got a couple of very interesting responses to this that we thought we'd share. So, first up is reader B.C. in Alexandria, VA:
I have been following your site for most of this election year but hadn't previously commented. However, I happen to be looking at some data last night which aligned with your comments on the Senate gap, as that also stood out to me as unusual.
I took a comparative look at the 2016 states where there appeared to be a significant gap between Trump's polling around this time, and the Senate candidate. Unsurprisingly, most of the swing states were included, with the exception of Michigan (as you noted, no Senate race that year). I was using October 20 as a point of reference. I looked at the polling numbers for each presidential candidate vs their opponent, compared this to the final percentages that they got on Election Night per Wikipedia, and noted gaps between the candidates on both sides.
I noticed a recurring pattern: Arizona, Florida, Georgia, Ohio, and Pennsylvania all featured races where (1) The Republican Senate candidate was leading while Trump was trailing; and (2) There was a significant gap (+4 to +10) between senator and Trump. In all five cases, both ended up winning, and Trump not only improved substantially, but ended up with a number within 1% of where the Senate candidate had been polling. This was the case regardless of whether the Democratic Senate candidate improved by roughly the same amount as the Republican (Florida, Pennsylvania), matched their polling (Arizona), or did worse than their polling (Georgia, Ohio). In those states, Hillary Clinton improved on her polling by 4 (Arizona), improved by 1 (Florida, Ohio, Pennsylvania) or matched her polling (Georgia). In all of these states except Florida, Clinton was within a point or two of the polling for the Senate candidate, which was also the case for Nevada and North Carolina.
Given this was the election where the supposed "Quiet Trump Vote" effect was in place, this trend seems suggestive, if not definitive. In states with strong Senate candidates, Republican voters who were vocally in support of their candidates but less openly in support of Trump in the home stretch seem to have consistently come around to Trump, regardless of how the margins moved otherwise.
How does this compare to the current election? At first it looks murky. As your data shows, Harris is within 2% of the Democratic candidate in every uneven state except Arizona, while Trump is running 3%+ ahead in every state except Texas and Wisconsin. Additionally, you don't seem to have mentioned Florida, where Trump is running 4% ahead of Sen. Rick Scott (R-FL) while Kamala Harris is 3% behind Rep. Debbie Mucarsel-Powell (D-FL). This would suggest that Harris is in Clinton's position, with polling more or less accurately reflecting what she's likely to get, rather than having a large number of supporters who are lukewarm about her that she needs to bring home. Wouldn't that leave the presidential races at a nailbiter margin, with potential for the Senate candidates to collect stray Trump supporters?
Actually, I think there's a strong chance the 2016 scenario is reversed. The social pressure most likely responsible for the "Quiet Trump Voter" phenomenon in 2016 was a negative bias against saying you were supporting Trump, because he wasn't viewed as serious, or held opinions offensive even among Republican voters. It does not appear to have been correspondingly positive towards Clinton, who matched her polling closely. However, the possible "Quiet Harris Voter" equivalent doesn't seem likely to be a negative bias against supporting Harris, who again seems to be capturing what you'd expect from her party. Instead, it is a positive bias towards saying you are voting for Trump, an affirmative requirement to not only support him, but do so openly. In other words, the people who say they are voting for Harris probably really are voting for Harris, but some of the voters who say they are voting for Trump might not be telling the truth.
It is almost incontrovertible that Trump is the defining figure of the current Republican Party, despite being out of office. This is evident in the way other current Republican politicians have been forced to bend to pressure from Trump, both legislatively in terms of pursuing his goals and also in terms of lining up behind him publicly despite having in the past been vocal critics (Sen. Lindsey Graham, R-SC; primary opponents Gov. Ron DeSantis, R-FL, and Nikki Haley; and his own VP J.D. Vance being prime examples). More tellingly, we can see what happens when people within the Republican sphere break ranks. Darlings of MAGA social media such as Kyle Rittenhouse and Joe Rogan, upon wavering in support of Trump, were immediately barraged with attacks from the right. Both walked back their statements quickly. One would expect that everyday voters who don't have their own cachet would have even less defense against such attacks, at least in their personal lives, if not on the same public scale. It strongly behooves people to maintain ranks publicly.
On the other hand, no other candidates appear likely to have their numbers goosed by this same enforced loyalty. The majority of the other Republican candidates are not incumbents who have accumulated support by default from their party members, and the ones who are (Sen. Ted Cruz, R-TX, and Scott) have among the lowest approval ratings in the Senate, with Cruz being among the bottom 10, as well as being famously despised by many of his colleagues. There is very little impetus for a voter to vocally claim support for Ted Cruz that they are actually questioning. Among the Democrats in these races, only Sen. Tammy Baldwin (D-WI) ranks among the least approved (more on this later).
So what happens if a reverse-2016 occurs, with Trump being dragged down to the level of his Senate candidates, as he was carried up previously, while Harris remains steady or gains a bit? Unsurprisingly, a complete blowout, with Trump losing every state you listed, and quite likely North Carolina as well, if Mark Robinson is similarly representative. Only Florida and Texas would remain red, and even they would be vulnerable to a scenario where Harris improved on her numbers as much as Clinton did in Arizona. Even if Trump was only brought down by half his current gap, he would still lose all of the same states. Even if only a few states played out this way it would swing the election. Swing Nevada, Arizona, and North Carolina, where the gap is largest (aside from Minnesota, which seems hard to call a swing state), and Trump has to win all of the others to win (winning all but Wisconsin would end in a tie).
And it's funny Wisconsin should come up there. Turning back briefly to the three other 2016 states in question, none of them are really in line with the main trend. In both North Carolina and Nevada, all four candidates in the state were polling within a few points of each other, and the final results had both Senator and Trump move by about the same, with North Carolina's more Republican-favoring population bringing both Trump and Sen. Richard Burr (R-NC) up to comfortable leads, and Trump actually beating would-be senator Joe Heck (R) by a bit, but not quite enough to beat Clinton. Wisconsin though, was a strange one. Sen. Ron Johnson (R-WI) went from being by far the furthest behind of the Senate candidates to winning fairly comfortably, and Trump likewise going from his worst polling to a narrow win. Presumably there are other significant factors at play there. But it does serve as a reminder that close polling doesn't guarantee victory, and even significant deficits can be reversed.
A funny last note: As I mentioned, Tammy Baldwin is the only Democratic Senator to be in the bottom 10 approval ratings (per Morning Consult). However, Ron Johnson was even worse, with only Mitch McConnell (R-KY), Bob Menendez (D-NJ), and Susan Collins (R-ME) being further underwater. McConnell is in the lose-lose situation of being disliked for both his successes and failures, and Menendez has already resigned. As for Collins... well, in the case where Harris wins but faces a 51-49 Senate, she could have some very interesting questions to ask herself.
And here is reader J.S. in Minneapolis, MN:
I'm curious whether some of the president-Senate gaps we see in the polling and election data come from under-voting or under-responding (i.e., declining to make a choice for either president or Senate), either instead of, or in addition to, ticket-splitting. So I spent some time analyzing the data you shared (21 observations). I found some possible evidence of this, but I think we'd need to look at individual-level data rather than aggregate data to be sure.
The following chart shows the gap between the Republican presidential candidate and the Republican Senate candidate on the Y axis, plotted against the difference in undecided voters between both major presidential candidates and Senate candidates on the X axis (calculated based on the total Rep + Dem president percentage minus Rep + Dem Senate percentage).
There does appear to be a correlation between the observed gap between Republican president and Senate candidates and the gap between the sum total of president and Senate candidates. In other words, the more the Republican presidential candidate is running ahead of the Republican Senate candidate, we see more undecideds/undervoting for Senate than president overall. Conversely, when the Republican senator is running ahead of the president, then we see more undecideds/undervoting for president than Senate. The slope (0.86) is statistically significant at p<0.01, and the R-sq is not trivial (0.58), though aggregate models and a small sample size require some caution.
This kind of makes sense: Think of low-engagement Trump voters showing up to vote for Trump and nobody/nothing else. But what's interesting is that we do NOT see the same pattern for Democratic candidates. The slope is 0.14 and not significant, and the R-sq is 0.04. It's plausible that this type of low information, president-only voter doesn't exist (at least not in the same numbers) on the Democratic side.
Thanks, J.S. and B.C.!
Although the two items are looking at different questions, they reach some similar conclusions: (1) That the wonkiness in the data appears to be on the Republican side, and not the Democratic side, and (2) it's possible that in some races, or many races, Donald Trump's support is not as strong as it appears. (Z)