Close states through the years
With decision time getting tantalizingly close for the presidential election, the seven states that we rate as Toss-ups have dominated political discourse—and our own internal thinking—to a perhaps unhealthy degree.
Wednesday night, David Plouffe, a veteran Obama strategist who is now advising the Harris campaign, said that all 7 Toss-up states are on track to be decided by a percentage point or less. At this point, we’d truly not be surprised if Plouffe’s prediction comes to pass. However, we did want to look back through history to see how common it was to have so many states decided by such narrow margins.
Table 1 considers presidential elections going back to 1976. We were a bit more liberal with our criteria, as we looked at the number of states that were decided by 2 percentage points or less each year. Part of this has to do with the small sample size of states that were decided by less than a point: of those elections, 2000 had the most states, 5.
Table 1: States decided by 2 percentage points or less, 1976-2020
Compared to Barack Obama’s elections, the Trump era has featured more marginal states: in 2012, Florida, with 29 electoral votes, was the sole state that ended up on Table 1. One advantage Democrats had during the Obama times was that the tipping point state (TPS on Table 1, the far-right column), voted to the left of the national popular vote in both elections: It was Colorado in both 2008 and 2012, which Obama won by 9 and 5 points, respectively. In 2016 and 2020, Wisconsin, the tipping point state, was decided by less than 1 point—and could very well be again. Ron Brownstein, writing for the Atlantic, noted our research on presidential “tipping points” and suggested the possibility that Trump may not have as big of an advantage in the Electoral College this year as he has had in the past.
As obvious as this may sound, Joe Biden was able to win in 2020 by reversing Trump’s advantage in the close states. In 2016, Trump won 4 of the closest states, representing 75 electoral votes, while Hillary Clinton took just 2, worth 14 electoral votes—we’d also note that Clinton’s states, Minnesota and New Hampshire, saw notable snaps back to Democrats in 2020, and Harris would be having a potentially catastrophic night if either were very close.
While presidential winners since 2008 have claimed a majority of the close states, before then, this was usually not the case. For instance, in 1984 and 1988, years when Republicans won comfortably, they built up such an advantage across the board that they could easily afford to lose the sole close states (Minnesota in 1984 and Washington state in 1988). Conversely, competitive races have produced situations where presidential losers claim more close states. Despite his narrow loss to Jimmy Carter, Gerald Ford’s relative strength on Table 1 stood out to us: 11 states were decided by 2 points or less in 1976, the most of any year we considered, and he took 8 of them. Four years later, the marginal states again broke decisively against Carter—he took 1 of the 7—although the roster of states was important: most of the states that Ronald Reagan carried narrowly that year were in the South and went heavily for Carter in 1976.
If we went back and redid this table after next week’s election, we’d be surprised if the tipping point state, whatever it ends up being, isn’t decided by less than 2 points. Having 3 consecutive elections where the tipping point state was so close would certainly be a historical anomaly, at least in the modern era, but it is another metric that speaks to the competitive nature of national elections these days.
Quick hits with days to go
— Michigan was Joe Biden’s best state of the 7 key swing states and North Carolina was Donald Trump’s best. Yet in another sign of the overall competitiveness of this race, neither Harris nor Trump have obviously put away what could or should be their lowest-hanging fruit of the key states, as best we can tell. One thing we were curious about in North Carolina was whether there would be any obvious sign of early voting turnout problems in western North Carolina on account of Hurricane Helene. Take a look at this turnout map from Catawba College Prof. Michael Bitzer if you’re curious—turnout in the area did not look markedly different than other parts of the state.
— A possible positive sign for Republicans is that in the midst of the close and competitive presidential election, the House generic ballot polling has moved a little bit their way. These are the polls that ask voters, typically, whether they support a Democrat or a Republican in their local House race. Different averages show different results: RealClearPolitics now has Republicans up about a point, while FiveThirtyEight has it roughly tied, but both have shown a little movement to Republicans lately. We are not sure what, if anything, to make of this, although if Republicans end up outperforming expectations in House and Senate races, perhaps this will have been an indication of it. It is also possible, from a House perspective, for one side to win the House majority without winning a plurality of the votes cast: Democrats did so in 1942 and Republicans did so in 1996 and 2012. The House popular vote, once fully calculated, should also be adjusted to account for uncontested races or districts where two members of the same party face off in the general election, as sometimes happens in top-two primary states like California and Washington. This will be something to look at once all the results are certified.
— Speaking of the House, there has been a smattering of nonpartisan polling in certain districts. Probably the best result for Democrats was that Rep. Matt Cartwright (D, PA-8) was up 50%-43% on businessman Rob Bresnahan (R) in his Trump-won, northeast Pennsylvania district in an Inside Elections/Noble Predictive Insights poll. Trump was up 49%-46% there on Harris, commensurate with Trump’s 2020 margin—that would be a good result for Harris if it held, because this covers Biden’s home turf of Scranton so some level of falloff might be expected there with the Biden to Harris shift. Emerson College found close races in some other places: Tom Barrett (R) led by 2 points in MI-7, an open seat Republicans view as one of their best pickup opportunities, while Rep. Yadira Caraveo (D, CO-8) led by 2 points herself in another Toss-up race. One other poll that cuts against a recent rating change we made was that Rudy Salas (D) led Rep. David Valadao (R, CA-22) by 2 points; we would still be mildly surprised if Valadao lost. Broadly speaking, we would say these polls are better for Democrats than the aforementioned generic ballot trend might indicate. Adam Carlson, a former Democratic pollster who has been doing solid work aggregating polls this cycle, recently took a look at the presidential numbers in House districts with publicly-released nonpartisan polls and found an average presidential swing of a little under 1.5 points to the right from the 2020 actual district-level figures to what the polls reported. That suggests a political environment a little less blue than 2020 but fairly similar—in other words, probably better for Democrats than what the generic ballot movement has shown, although the comparison is not quite the same (the generic ballot measures House voting intention, while Carlson’s comparison focused on presidential performance at the House level).
— There were a couple of poll results that caught our eye in the Industrial North presidential/Senate races. The general trend throughout the cycle has been that Democratic Senate candidates have run ahead of the Democratic presidential candidates in Michigan, Pennsylvania, and Wisconsin, if not necessarily in vote share than at least in margin over their Republican rivals. But a couple of polls found differently here at the end of the campaign. A Susquehanna poll of Michigan found Harris up 52%-47%—a great poll for Harris there compared to some others—but Rep. Elissa Slotkin (D, MI-7) was up just 48%-47% on former Rep. Mike Rogers, her GOP rival. Meanwhile, a CNN/SSRS poll showed another strong showing for Harris in Wisconsin, up 51%-45% on Trump, but Sen. Tammy Baldwin (D-WI) was up just 49%-47% on businessman Eric Hovde (R). To be clear, these are outliers, but they do remind us of something from 2016. The final Marquette University Law School poll of Wisconsin that year had Hillary Clinton up 46%-40% on Trump, but former Sen. Russ Feingold (D) up just 45%-44% on Sen. Ron Johnson (R) as his race tightened down the stretch. The poll didn’t pick up on either Trump or Johnson winning, as each ultimately did, but the poll did suggest that Johnson was doing better in the Senate race than Trump in the presidential, which is what happened. As an aside, this year’s final Marquette poll had Harris up 50%-49% and Baldwin up 51%-49%, which better reflects other polling this year. The lion’s share of the data suggests that, to the extent there’s a difference between the presidential and the Senate races in these 3 battlegrounds, Democrats are doing a little better at the Senate level than the presidential, although down the stretch Baldwin’s showing has come to most closely resemble the presidential polling. Could one of these Republican Senate candidates in the Industrial North states do better than Trump? That would be a surprise, but we at least wanted to note the possibility, as illustrated by these polls.