The hopes have been building among Democrats. What if — maybe, just maybe — this year’s midterm elections won’t be so bad for their party after all?

The weight of history, polling numbers, and last November’s election results all seemed to spell doom for Biden’s party in the midterms. But in recent months — particularly since the Supreme Court’s Dobbs decision eliminating federal abortion rights protections — there’s been a shift.

Recent special elections have been encouraging for Democrats, most recently with their pickup of Alaska’s House seat. The lead in generic ballot polling that Republicans had all year vanished last month, and current polling averages show about a tie. And if Democrats win all the Senate races where they’re currently leading polls, they’d retain control of the chamber (though some of those leads are quite small, and some states haven’t been polled very often).

It’s important to keep things in perspective. Polls are not showing a blue wave — instead, they suggest Democrats are underdogs to retain the House and that the Senate could go either way (FiveThirtyEight’s forecast gives the GOP about a one-in-three chance of winning it). Current indicators now seem to point to a close midterm contest rather than a blowout defeat of the type suffered by Donald Trump’s GOP in 2018, Barack Obama’s Democrats in 2014 and 2010, and George W. Bush’s GOP in 2006.

But is this a true change in the political winds — or will Republicans end up with a solid victory in the end after all?

There are two basic schools of thought about what’s going on here.

One theory is that Democrats’ improvement is real, largely due to an improved news environment for the party, and likely to last. In 2021, the Democratic base was demobilized, Republicans were fired up to vote, and swing voters were tipping toward the GOP. Then, the Dobbs decision and former president Trump’s return to the headlines has energized the Democratic base, too, as well as putting swing voters’ focus back on Republican extremism. Meanwhile, falling gas prices have put some of voters’ economic fears at ease. So rather than a blowout, this looks like an election where the parties are pretty evenly matched.

The second theory is that this won’t last — either Democrats’ recent advantage will prove transient or it doesn’t exist at all. That is to say, Republicans’ poll results could improve later, as happened in Virginia and New Jersey in October 2021, because of either typical midterm trends or a shift in the news environment. Alternatively, the polls could just be wrong — they could be systematically underestimating the GOP’s strength, as happened in 2016 and 2020 (and in certain regions in 2018, too). As for Democrats’ recent good special election results, they might not be nationally representative, or they might not be matched in the comparatively higher-turnout midterms.

The case that Democrats really have improved

The history is clear that the president’s party usually does poorly in midterm elections, as I wrote last year. There are different ways to measure exactly how poorly — change in House seats, national House vote margin, Senate seats, and governorships. But in each metric, losses for the president’s party are common, and large losses are more common than even small gains.

Andrew Prokop/Vox

Why does this so often happen? Midterms may be inherently demobilizing to many of the incumbent president’s supporters precisely because he’s not on the ballot — they feel less threatened because they know he’ll still be in office no matter how the midterms turn out, and therefore are less motivated to vote. Political scientists have also put forward the “thermostatic” public opinion model, suggesting that swing voters tend to swing against the incumbent party, thinking that the country has been moved too far to either the left or right.

There have been notable midterm non-blowouts, though. The 1998 elections were conducted amid a roaring economy and congressional Republicans’ unpopular effort to impeach President Bill Clinton, and they were basically a draw. The 2002 midterms were the first since 9/11, President George W. Bush was still quite popular, and the GOP did well.

So is there reason to think Biden’s Democrats might defy the trend and avoid a smashing midterm defeat — becoming one of these exceptions?

The Dobbs decision, which wiped out a legal status quo that had existed for half a century, may well fit the bill. The conservative Supreme Court justices’ elimination of federal abortion rights protections is a rare example of a dramatic policy shift clearly opposed by the incumbent president. Millions more Americans confronted the possibility that if they or someone in their family should need an abortion, they could be blocked from getting it by the government. Perhaps it has mobilized Democratic base voters who’d otherwise tune out for the midterms, and convinced swing voters that Republicans have moved the country too far to the right.

There are other ways the news environment has improved for Democrats. The news in the second half of 2021 and the beginning of 2022 was a seemingly endless parade of horribles for them: Covid resurging with new variants, inflation soaring and real incomes falling, the messy withdrawal from Afghanistan, a stalled legislative agenda, and the unpopular former president Trump mostly out of the headlines.

In the past few months, things have been different. Covid has receded from headlines, and mask mandates have been lifted even in blue areas. Gas prices have fallen substantially since the summer peak (though inflation may not yet have been stopped). Biden and Democrats passed the Inflation Reduction Act, a scaled-back bill that nonetheless made good on some of their campaign pledges. And the January 6 hearings and the FBI search of Mar-a-Lago returned Trump and his problems with the law to the headlines.

Sure, not everything looks rosy — President Joe Biden’s approval rating has improved by about 5 points since July, but it’s still under 43 percent, which is historically rather low. But as Amy Walter writes at the Cook Political Report, there’s a case that this is the wrong metric to focus on. Some polls show that a substantial portion of Biden disapprovers dislike Republicans even more and say they plan to vote Democratic in November. Furthermore, the post-Trump Democratic coalition, which relies heavily on college-educated voters who are more likely to turn out, may now be well-optimized for midterm elections.

The argument goes that all this can be seen in Democrats’ improved poll numbers and, even more tellingly, in special election results. A Kansas referendum on abortion rights in August resulted in a blowout win for the pro-choice side. Post-Dobbs special House elections in Nebraska, Minnesota, New York, and Alaska all showed marked strength for Democrats, too. All of this is a case for why the midterms may well be something closer to a draw, or a tight and closely fought contest, than a GOP wave.

The case(s) that Republicans will emerge triumphant

But there are also reasons the above theory (and the current polls) might turn out to be wrong by Election Day.

For one, there could be late movement in Republicans’ favor. So maybe the current polls are pretty accurate, but they’ll later change, and the GOP will get an advantage in the campaign’s final weeks. Late poll movement is a real phenomenon, as Hillary Clinton could tell you — she led solidly through October and then collapsed after James Comey’s infamous announcement that the FBI had reopened the investigation into her emails.

We don’t have to reach too far back in time for other examples. In 2021, Terry McAuliffe (D) led Glenn Youngkin (R) in public polling of the Virginia governor’s race all the way up to the last week of October, when Youngkin took the lead. It’s difficult to disentangle why, exactly, Youngkin surged late. It could have been due to contingent factors (like a McAuliffe campaign gaffe) or broader ones (like a tendency for these elections to break late against the incumbent president’s party). But it did happen. The polls in the end were quite accurate in Virginia; it’s just that they had changed from where they were in even mid-October.

As mentioned above, Democrats have been benefiting from a notably favorable news environment over the past few months, in stark contrast to how 2021 ended and 2022 began for them. We don’t know what the campaign’s remaining weeks will hold, but unfavorable news developments in the economy (like Tuesday’s worse-than-expected inflation announcement) or on any number of topics could certainly hurt them. In a vacuum, it is possible that late movement could benefit either Democrats or Republicans, but the weight of midterm history could be read to suggest it’s more likely to hurt the incumbent president’s party.

Additionally, much has been written about the Democrats’ advantage in ad spending in key Senate races so far, but one reason that exists is that Republicans have been saving their comparatively more limited resources to spend on ads later. So it’s possible that once Republicans do unleash their negative ads, the narrow leads some Democratic candidates currently have will disappear with late movement.

So that’s how the polls could change. But they could also just be wrong — again.

The New York Times’s Nate Cohn broached this possibility Monday, writing that a “warning sign” is flashing in Senate polling. Specifically, in states where polls dramatically underestimated Republicans in 2020 — like Wisconsin, Ohio, and Florida — polls show 2022 Democratic candidates doing surprisingly well. It’s possible, then, that those polls are mirages.

This famously happened in the 2016 and 2020 presidential years, but even in the 2018 midterms, this problem occurred in many states (though not everywhere). Since the Trump realignment, pollsters have often struggled with the problem of underestimating GOP support. Initially, some theorized that the polls were underweighting non-college-educated voters, but the problem persisted despite efforts to correct it.

It’s possible that this time really is different. It’s useful to be mindful of the polling problems of the past, but that doesn’t necessarily mean they’ll repeat this year. Still, Democrats have been burned enough by rosy polls that some caution is justifiable. The likelihood of a red wave may have decreased, but it can’t be counted out just yet.