James Lawler Duggan / Reuters

House Democrats are on recess and back in their districts this week, gauging how their constituents feel about impeachment. Public opinion has moved toward support of the impeachment inquiry, but many Democratic strategists remain especially concerned that the 31 House Democrats who represent districts that backed President Donald Trump in 2016 could face a backlash if they ultimately vote to impeach him.

That’s led some to privately contend that if House Democrats eventually bring articles of impeachment to a floor vote, the party leadership should not push members from swing districts to support it. But electoral history of the past two decades suggest that these vulnerable Democrats may be safer hanging with their party than trying to maneuver separately.

The reason is that it is far from clear that swing-district members can insulate themselves from the broader public reaction to impeachment, however they vote individually. On other highly contentious votes in recent years—such as the assault-weapons ban in 1994, the passage of the Affordable Care Act in 2010, and the attempted repeal of the ACA in 2017—the backlash against those decisions in the next election washed away both representatives who voted with their party and those who didn’t.

That suggests to some analysts that even House Democrats in vulnerable seats may be better off politically if they vote yes on impeachment and help their party send the electorate a united message than if they hedge their bets by voting no, at the price of signaling to voters more disagreement over the decision.

“I think it’s very hard to separate yourself from the national vote once it’s cast,” Dan Sena, who served as executive director of the Democratic Congressional Campaign Committee during the 2018 election, told me. “The more unified the Democratic Party is, the more unified the message will be.”

Since Nancy Pelosi announced that Democrats would officially start impeachment proceedings two weeks ago, a number of polls have shown a majority support for the House decision to begin an impeachment inquiry—though few have shown majority support yet for impeaching and removing Trump from office. This week, polls from NBC/The Wall Street Journal and Quinnipiac University showed a plurality of voters opposed to removing Trump, while an ABC/Washington Post survey found a 49 percent plurality in support. However, a Fox News poll released yesterday did show support for Trump’s removal reaching a slim majority of 51 percent, an ominous sign for the president.

Support for Trump’s removal could rise further as the inquiry proceeds—as I’ve noted, while polls show that virtually all voters who approve of Trump’s job performance already oppose impeachment, there’s more room for support of impeachment to grow among voters who disapprove of him overall. Still, unless new information somehow cracks Trump’s hold on his base, impeachment likely will remain a nerve-racking political vote for Democrats in the districts that Trump carried in 2016.

That raises a crucial question: If an impeachment vote reaches the full House, should marginal Democrats best position themselves for 2020 by separating themselves from the party by voting no, or helping the party send a clearer message by voting yes?

The precedent of other polarizing high-profile votes in recent years doesn’t offer a decisive answer. But the evidence leans toward the conclusion that it is growing more difficult for members of Congress to isolate themselves from their party’s actions, regardless of how they individually vote. Political-science research shows that for more and more voters, congressional elections are becoming more parliamentary, meaning they are less a choice between two candidates than a decision about which of the two parties they want to control the majority and set the country’s direction. That’s seen in the dwindling share of voters who back one party for president and the other for Congress; it’s now routine for the president’s party to win more than 90 percent of the voters who approve of his job performance in House and Senate races, and to lose more than 90 percent of the voters who disapprove.

“What’s happened is, over the last decade or so, these elections are party-branding exercises now,” says Tom Davis, a Washington-based lobbyist who was formerly a Republican representative from Virginia and chair of the National Republican Congressional Committee. “What you find today is members are constantly hostage to their leadership. Whatever they do as individuals in the eyes of their constituents doesn’t matter, because they have given the gavel and control [to one side].”

Still, the evidence remains mixed about the extent to which dissenting from a party consensus can shield members of Congress from the overall reaction to those decisions.

In recent history, one of the most prominent examples of members of Congress trying to shield themselves from the party consensus came with the bitterly contested House vote in 1994 to impose a ban on assault weapons. Seventy-seven House Democrats, mostly from rural and southern districts, voted against it. But many of them lost anyway in the GOP sweep of the midterm elections that fall.

Gary C. Jacobson, a UC San Diego political scientist who specializes in congressional politics, told me research on the 1994 election has found that House Democrats in Republican-leaning districts who opposed the assault ban actually had a higher survival rate than those who supported it. But he says that such attempts at political separation are experiencing diminishing returns. His research has shown that Republicans in competitive districts who distanced themselves more from Trump in their voting record did not fare any better in the 2018 election than those who more consistently voted with him. “For this most recent election … it was all national,” Jacobson said. “The predictor of your vote in the district was basically how Trump had done in 2016. There was very little opportunity to differentiate yourself.”

The 2017 vote on whether to repeal the ACA, perhaps the highest-profile House decision of Trump’s first two years, mostly supports that conclusion—but also offers a caveat. In the 2018 midterms, Democrats pounded Republicans over the ACA repeal, helping them win control of the House majority. Voting against the repeal offered some defense against those attacks for vulnerable Republicans, but hardly an impenetrable one. In 2018, Democrats captured half of the seats held by the 20 Republicans who voted against repeal: They defeated four incumbents who voted no, and captured the open seats left behind by six other Republicans who retired, almost all in districts where Trump was unpopular.

But the only three House Republicans representing Clinton-won districts who survived 2018 voted against the ACA repeal: John Katko of New York, Brian Fitzpatrick of Pennsylvania, and Will Hurd of Texas (who is retiring next year in a seat where Democrats are now favored).

A GOP consultant who helped one of the three Republicans get reelected, and who spoke on the condition of anonymity to discuss internal campaign strategy, told me that opposing the ACA was a crucial piece of “supporting evidence” that the incumbent used to portray himself as an independent figure in a district dubious of Trump. “We were able to grab those snippets and make it a hallmark of the campaign and survive,” the consultant said.

The GOP wave in the 2010 midterm elections reveals a similarly nuanced trend. In a study after the election, the University of Denver political scientist Seth Masket and a colleague concluded that voting for or against the Democrats’ highly polarizing “cap-and-trade” climate bill didn’t affect vulnerable Democrats’ chances of withstanding the Republican wave, yet Democrats in marginal seats who opposed the ACA were more likely to survive than those who backed it. But once again, that defense was hardly impenetrable: Of the 34 Democrats who opposed the ACA and ran for reelection, half still lost.

These precedents—the assault ban in 1994, the cap-and-trade bill in 2009, the passage of the ACA in 2010, and the attempted ACA repeal in 2017—are all examples of bills that faced a genuine backlash among certain swaths of the electorate.

The electoral reaction to the GOP impeachment of Bill Clinton in 1998 represented the opposite case: Though Clinton’s impeachment was unpopular in polls, the evidence is strong that it was not a huge issue for voters. Despite the unpopularity of impeachment, the GOP didn’t lose control of the House in either the 1998 or 2000 elections; over those two elections, only nine Republican incumbents were defeated in their reelection bids. It’s true those losses were concentrated among Republicans who held districts that had backed Clinton in 1996, where impeachment presumably drew the least support. But the majority of the 91 House Republicans from Clinton districts who sought reelection won it, even after voting to impeach him.

In cases when a genuine wave of backlash developed, as in the vote to repeal the ACA, dissenting from the party offered representatives no guarantee of survival. The Clinton impeachment example, in turn, showed that sticking with the party didn’t ensure defeat when a big wave of backlash failed to materialize. In both cases, the key electoral dynamic was the size of the wave, not how members positioned themselves to weather it.

Jacobson told me this long-term trend toward more parliamentary-style voting means that House Democrats wavering over impeachment may be better off sending voters a strong signal by contributing to the party’s nearly unanimous vote than by opposing it, which would help Trump and the GOP portray impeachment as a factional decision driven only by the Democratic left.

“I think they are better off trying to send a unified message,” he said. “It may be not true in a handful of districts … but they want to make a statement this is not just partisan politics; it is upholding the Constitution. And under those circumstances, unity is a much more powerful statement.”

Sena told me that for House members in marginal seats, how the party handles impeachment may matter more than whether Trump is impeached. “I think the process may matter more than the actual result, because my suspicion is the electorate will go back to where they have been” on Trump after the impeachment debate ends, he said.

Masket, the University of Denver political scientist, cautions that the ACA precedent signals that for some of the House Democrats in Trump districts, voting no on impeachment may still be the safest course. “I think that same logic holds,” he told me. “It’s not necessarily an exact mirror of the ACA, but there still are moderate districts, and there still are voters who can draw a distinction between a conservative rural Democrat and, say, Nancy Pelosi.”

Public support for Trump’s removal, Masket said, could reach a point where the risk of voting for impeachment is minimal even in most of the Trump districts. But he notes that House Democrats are unlikely to obtain the one thing that could make it easiest to defend an impeachment vote in such places: support from enough House Republicans to give impeachment a bipartisan veneer. “I think if there are Republicans going along with this, that gives those marginal Democrats a lot more cover,” he told me.

Absent that cover, swing-district Democrats are heading toward difficult choices in the weeks ahead. Most of them appear on track to vote for articles of impeachment; all but seven have endorsed the impeachment inquiry. But whether they support impeachment or not, their fate in 2020 may hinge less on their individual votes than on the country’s verdict on the overall impeachment process—and even more so on how it assesses Trump’s term in office.

Swing-district Democrats will likely try to buck their party, but the political calculus on whether they survive in 2020 may be a lot simpler than they are likely to believe. As the Never-Trump Republican Bill Kristol put it, vulnerable Democrats “are going to get reelected if it’s a Democratic year, and if it’s a really good Trump year, they are going to lose.”

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.