In Washington on Sunday night, the popcorn was popped, the pitchers of election-themed cocktails given a final stir. And then a hush fell over the city as the streets emptied and the professional political class gathered huddled around screens to find out whether Donald Trump would somehow rescue his improbable campaign, or whether he would pound the final nails into its coffin after a brutal weekend of scandalous recordings and cascading Republican disendorsements.

The consensus going in was that Trump would self-destruct. But he had two saviors: moderators prevented him from rambling and harming himself as much as he otherwise probably would have, and the questions about his appalling remarks came early in the debate. Psychological research suggests that people tend to disproportionately judge events by how they ended.

The consensus in Washington was that Trump had “stopped the bleeding.” But 12 hours later the hemorrhaging seemed to have restarted. NBC and the Wall Street Journal released the results of a poll taken before the debate, but after the release of the bombshell tape that had sent Trump’s campaign into its weekend death spiral. In a four-way race, it showed Clinton at 46 and Trump at 35, a five-point drop from the previous poll. If you just looked at Clinton and Trump together, she was leading by 14.

The question now is whether the defections restart. If Trump’s debate achievement was to arrest his slide at 35 percent, then it makes sense for Republicans to start jumping into the lifeboats.

Here’s the basic calculation that Republicans face re-election now face: unendorse Trump, and see his base stay home, or endorse, and lose swing voters who are appalled by his remarks.

At this juncture I think of Republican voters as falling into five broad categories, each with a different propensity for retaliation against Republican candidates who oppose Trump:

1. Die-hard Trump supporters: These people wanted Trump, and only Trump, in the primaries. They hate the establishment. They viewed Trump as a weapon against the establishment, and will support sitting politicians only to the extent that those folks go along with Trump.

Propensity for retaliation: Absolute. Incumbents who oppose Trump can expect these voters to refuse to vote for them, or even to vote for their opponent.

2. Trump primary voters: These folks voted for Trump in the primary but are not engaged in an all-out war on the establishment of their party. They liked the fact that he’s an outsider. They admire his business acumen. They would rather listen to his unpolished style than the boring, focus-grouped talking points of a normal politician.

Propensity for retaliation: Medium-high. These people would have voted for another Republican, but they prefer Trump, and they will view an unendorsement as an attempt by leadership to thwart the will of the voters. However, some unknown number of these voters will have been offended by the tape, and will probably not retaliate against Trump defectors.

3. Party loyalists: These folks probably voted for someone else in the primary, but do not feel the same horrified revulsion against him that #NeverTrump voters do. They will pull the lever for anyone with an (R) after their name. Their biggest concern is maximizing the party’s power.

Propensity for retaliation: Medium-low. They also don’t like the idea of the party ignoring the voters. On the other hand, they are horrified by the idea of giving a Senate or House seat to a Democrat. The incumbents‘ biggest risk with these people is that they get demoralized and stay home.

4. Squishy Republicans: These folks usually vote Republican, but may not be registered as a Republican, and may occasionally drift over to the other side of the ballot. They are not particularly interested in an attempt to remake the party into a more nationalist, populist entity, and are not going to participate in the Trumpista revolt. They will also find it psychologically easier to support Clinton than the aforementioned groups if they are horrified by the tape.

Propensity for retaliation: Low. They might stay home. They might vote for Clinton. But they are probably not going to get involved in strategically placing their congressional votes in order to serve some grand vision of the party’s future.

5. #NeverTrump: Uncertain in number, but high in passion, these people will vote Johnson or McMullin or write in their Aunt Agnes before they will ever pull the lever for Trump. They tend to be loyal Republicans, but some of them have cancelled their registrations over Trump.

Propensity for retaliation: None. They may phone bank for you if you call on Trump to step aside.

The problem is, no one knows exactly how big any of these groups is. Republican politicians essentially have to make a blind bet: Do I lose more swing voters by sticking with Trump, or more base voters by defecting from him?

If the results of the NBC/WSJ poll hold, they suggest that Trump has not only lost all swing voters, but is now cutting into the Republican base -- and not just #NeverTrumpers, because he’d already lost them. The next question is: Does swing voters’ animus toward Trump also affect their vote for Republicans in general?

There’s no way to know yet, but my guess would be that at least some downticket Republicans will lose if they stand with Trump. What’s on the tape is not spinnable, especially given prior accusations that he’s actually done things similar to what he talked about on tape. The public hasn’t heard as much about that yet. But by November, they will have.

By now there may be no way to win; no matter what they do, down-ticket Republicans may lose too many voters to win their election. But not every loss is created equal: Even the doomed can decide to die with honor.

Megan McArdle is a Bloomberg View columnist. –Ed.

By Megan McArdle
Bloomberg