Two weeks ago, I wrote about the survey data that showed fear, unity, and high presidential ratings in the immediate aftermath of the 9/11 terrorist attacks in 2001. The midterm five years later told a completely different story, marked by resurgent partisan division, unpopular wars abroad, and a very unpopular second-term president.
Analysis of the 2026 midterm environment often looks to 2018 as a point of comparison. It’s an understandable choice since the 2018 midterms also hinged on voters’ perceptions of a Donald Trump presidency. Recency bias also leads us to look to more recent events for insight. Certainly there are similarities between 2026 and 2018, but where we are 20 years later looks remarkably like 2006.
President George W. Bush’s 2006 approval ratings were essentially where Trump’s are now: in the toilet. Bush's ratings, still damaged by his administration’s mishandling of Hurricane Katrina in New Orleans, hovered in the mid-30s headed into the midterm elections of his second term. Bush was stuck in the mire of wars in Afghanistan and Iraq, with no signs of resolution three years after he had declared “Mission Accomplished” in the latter. The economy was not terrible yet, but after four years of steady increases, gas prices averaged $4.22 per gallon in 2026 dollars. That’s only slightly lower than the current average.
Bush’s deep unpopularity meant that Republicans did not want to campaign with him. Unlike Trump now, Bush mostly stayed off the campaign trail and on the sidelines in 2006. Democrats ran the table in the midterm elections, gaining Senate seats in Montana, Missouri, Ohio, Pennsylvania, Rhode Island, and Virginia to move the chamber from a 55-45 Republican majority to a 51-49 Democratic majority.
In the House, Democrats flipped 31 seats, shifting a 232-203 Republican majority to a 233-202 Democratic advantage—their first House majority since the Republican takeover under Newt Gingrich in 1994. They swept seats throughout the Midwest, Northeast, and southwest Texas.
At this juncture in 2018, Trump’s approval ratings had hovered for months within a narrow range of 41-43 percent. Gallup’s measures kept him mostly above the 40-percent mark. That is 4-5 points higher than his current ratings, which average 37-39 percent across various poll aggregators. According to Gallup, Bush’s approval ratings in 2006 were 33-38 percent.
Democrats did not win the Senate in 2018. This fact is easy to forget in retrospect, but Republicans netted two Senate seats by flipping North Dakota, Missouri, Indiana, and Florida. Democrats gained a seat in Nevada and one in Arizona. Most of the discourse focuses on the 40-seat Democratic gains in the House, well over the 23 they needed to take the majority. The popular vote across all House races favored Democrats by nearly 9 points.
The expected outcomes for 2026 look more like 2006. Democrats need to gain 4 seats to take over the Senate, and right now it looks doable. Notably, as in 2006, Democrats could take Senate seats they really shouldn’t be able to in any other year. The 2006 anomalies were Montana and Missouri; this year, Texas and Iowa are in play. No one is talking seriously about Republicans gaining seats, although New Hampshire and Michigan could slide to the right if there’s a systematic polling error that’s currently underestimating support for GOP candidates.
Democratic Senate candidates lead in polling across most battleground states, with surveys showing a strong chance of Sen. Jon Ossoff getting reelected in Georgia and former North Carolina Gov. Roy Cooper flipping the seat Sen. Thom Tillis is vacating.
That leaves seven races rated toss-ups by The Cook Political Report with Amy Walter. Of those, Democrats need to hold New Hampshire and Michigan, plus pick up three among Texas, Iowa, Ohio, Maine, and Alaska. It would be a tall order in any year—including this one—but most quantitative forecasts give Democrats a little more than 50 percent chance of pulling it off.
On the House side, one substantial change since 2006—and even since 2018—is that fewer seats are competitive. Democratic gains are unlikely to match 2018’s 40-seat increase because structural changes reduced competitiveness after the 2020 Census, then again through the mid-decade gerrymanders.
Cook rates only 21 districts as true toss-ups, with another 16 classified as “leaning” to one party or the other. If Democrats win all the toss-ups and “lean Democrat” districts currently held by Republicans, that is a gain of 18 seats. Winning all of the “lean Republican” districts would add another 5 seats for Democrats, inching upward to a 23-seat gain. It’s more than enough to win the majority, but by far fewer seats than were competitive in previous midterms.
The best reasonable outcomes for Democrats are Senate gains of +4-6 and House gains of +18-23. Indeed, this best-case scenario looks more like 2006 (+6 and +31) than 2018 (-2 and +40).
Contributing editor Natalie Jackson is founder and principal of Centerline Research and Strategy.
