One of the most intriguing questions going into next year’s Senate races is which party will be playing offense and which will be content to defend. History argues that Democrats will be on defense and Republicans on offense; exposure argues the opposite way, that Republicans will play defense while Democrats look to pick off seats. 

Almost everyone uses the post-World War II time frame when comparing midterm elections, but politics has changed a lot during that time. Straight-party voting is much more common, for instance. If we limit our sample to more recent times, exceptions to the rules start to become more common, as when the parties of President Clinton in 1998 and President George W. Bush in 2002 gained seats. 

For all its drawbacks, the era since World War II is indeed the best time frame for comparisons—as long as we focus only on elected presidents. Vice presidents stepping into the Oval Office, as Lyndon Johnson did after the assassination of President Kennedy and Gerald Ford did after President Nixon’s resignation, simply creates different dynamics. Among elected postwar presidents, the average loss in their first midterm has been three seats in the Senate and 22 seats in the House.

But it is also important to separate first-term midterm elections from those in a second term. While the difference in the House is negligible, (23 losses for the former and 20 seats for the latter), in the Senate it is massive. The first-term average loss by a president’s party is only one seat; in second-term midterms, that average swells to seven seats—a distinction with a real difference. 

Of course with the Senate currently split 50-50, any net loss for Democrats is fatal, so they have to beat the averages to hang on.  

There’s little if any ticket-splitting in presidential years. Readers will recall that 2016 was the first election since the advent of direct election of U.S. senators in which every single Senate race was won by the same party that carried that state in the presidential race. In 2020 it was all but one—Republican Sen. Susan Collins of Maine being the lone survivor in a state won by Joe Biden.

But of the 35 Senate races held in 2018, in the most recent midterm election, seven Democrats managed to win in states that Donald Trump had carried two years earlier: Kyrsten Sinema in Arizona, Debbie Stabenow in Michigan, Jon Tester in Montana, Sherrod Brown in Ohio, Bob Casey in Pennsylvania, Joe Manchin in West Virginia, and Tammy Baldwin in Wisconsin. Four Democratic incumbents lost in states Trump had won two years earlier: Bill Nelson in Florida, Joe Donnelly in Indiana, Claire McCaskill in Missouri, and Heidi Heitkamp in North Dakota. 

Republicans lost the only seat they held in a state where Hillary Clinton had prevailed, the Nevada seat held by Dean Heller. They also lost an open seat they were defending in Arizona, a state that had backed Trump in 2016 (although Arizona did flip to Biden in 2020 and Republicans lost another Senate seat there that night).

What about the previous midterm? In 2014, no Democrat won a state that Mitt Romney had carried two years before, and Republicans managed to win in two states that had voted to reelect President Obama: Colorado (Cory Gardner) and Iowa (Joni Ernst). Of course, the former lost reelection last year when the state voted Democrat for president yet again; the undertow was too great to hold the Senate seat. The Iowa that had voted for Obama in 2012 was gone, with Trump taking the Hawkeye State in both 2016 and 2020, and Joni Ernst winning reelection last year by a stronger-than-expected margin.

Let’s put history aside for a moment and look at exposure, or how many seats a party must defend. The GOP has 20 seats up to just 14 for Democrats—numbers which make the case that Republicans may end up on defense. Worse yet for Republicans, five of those 20 seats are open, whereas thus far Democrats have zero open seats to worry about. Over the last 20 years, 86 percent of Senate incumbents seeking reelection have won. Incumbents’ worst year since 2000 was a 78 percent win rate; the best, 91 percent. So you’d much rather defend a seat that’s occupied than one that’s not.

So will Democrats be playing offense or defense? At this point, the answer is probably “yes”—that it will be a knock-down, drag-out fight for every single seat, any one of which could tip or hold the Senate. How much fun is this going to be?

This article was originally published for the National Journal on March 16, 2021.

More from the Cook Political Report