Long before the rise of the polling industrial complex, voters still hungered to know the election results in advance.

One hundred years ago, in the run-up to the 1922 off-year elections, newspaper reporters offered predictions based on interviews with national party officials and local insiders. 

And they weren’t half bad.

An election-eve analysis in 1922 by the Chicago Tribune estimated that the Republican margin in the House would dwindle to 25 to 30 seats. In truth, the out-of-power Democrats gained 77 seats to cut the GOP majority to just 18. 

These days, of course, the politically obsessed devour every poll and hyperventilate if a candidate gains 1.8 percent among women who attended college but did not graduate. And if that poll is coupled with another survey that shows a gain of 0.7 percent among left-handed independents, then we have a certifiable major political trend.

But I do wonder if there is much clarity to be gained by constantly checking the forecasts from FiveThirtyEight, the polling averages from Real Clear Politics and the district-by-district ratings from the Cook Political Report. 

The truth is that this may be an off-year election in which the outcome is unknowable until Election Day. Or, given the speed that states like California and New York tally votes, control of the House may be unknowable until mid-November.

Six weeks before the election, it is easy to forget the weirdness of the 2022 political environment. 

Wednesday’s hearing of the Jan. 6 committee should serve as a reminder that this is the first off-year election since a violent mob ransacked the Capitol. And, despite Dr. Joseph R. Biden’s medical assurances that the pandemic is over, there is no way to reckon the lasting effects of COVID-19 on the national psyche. 

Finally, the news in normal off-year elections does not center around an election-denying, defeated ex-president with a penchant for hoarding top-secret documents at his golf club home. 

Against this backdrop, it is also difficult to know exactly how much weight to put on Biden’s approval ratings, which have bounced up to 43 percent in the Real Clear Politics averages. Adding to the murkiness, the generic ballot (asking voters what party they are backing for the House) is knotted at 45 percent, according to the website’s averages. 

A robust debate over the accuracy of polls adds a further note of humility to any election predictions. 

In 2016 and 2020, pollsters failed to fully adjust for the reluctance of Donald Trump partisans to answer survey questions. The result of this inadvertent Democratic tilt was not only an embarrassing failure to anticipate Hillary Clinton’s 2016 defeat, but also the wrongheaded belief that 2020 Senate races in states such as Maine and South Carolina were competitive rather than GOP romps. 

Have the pollsters improved their methodologies? Does the Trump factor disappear if the Mar-a-Lago souvenir hunter isn’t on the ballot? Can the polls showing a strong Democratic comeback be trusted?

These are, of course, questions that will only be answered by the voters in November. But good luck in deciphering the polling averages until we know for sure. 

All election predictions traditionally have been hedged by factoring in a wild-card allowance for an October surprise. 

And we have certainly had them. In 1962, America went to the brink of nuclear war with the Soviet Union just before the midterms. (The results: The Democrats gained four Senate seats and lost four House seats in an election that was mostly a wash.) 

And in 2016, the October surprises were thermonuclear. First came Donald Trump’s “Access Hollywood” tape. And then, 11 days before the election, FBI Director James Comey announced that he had reopened the Hillary Clinton email investigation.  

True surprises, of course, defy expectations. But at the moment, it is hard to divine an external event that will change the contours of the 2022 midterms.

While gasoline prices have stopped falling, there are no indications that they are likely to spike before the election. Voters are much more likely to gauge the severity of inflation based on the price at the pump rather than the final preelection number announced by the Bureau of Labor Statistics in October.  

The Federal Reserve is risking triggering a recession as it sharply raises interest rates to ward off inflation. But with the unemployment rate near a historic low at 3.7 percent, layoff notices may prove to be a 2024 issue, but they won’t affect the midterms. 

The world is more volatile than ever with Vladimir Putin desperately thrashing about to stave off defeat in Ukraine while Iranian women are displaying inspiring courage as they battle the theocracy. But it is very rare for foreign affairs to influence American elections unless the lives of U.S. soldiers are at stake. 

Probably, the biggest wild card is (surprise) Donald Trump.

While Trump’s fervent supporters remain impervious to reality or disillusionment, there is always the political danger for Republicans that a new revelation might test the loyalty of suburban voters who back the GOP for economic reasons. No issue is more explosive for Trump than his lack of a coherent reason why he kept nuclear documents in a storeroom protected by a hardware-store lock. 

In the end, we are probably going to be where we are today — looking at a tight set of midterms with no guarantee that past precedents or current polling are reliable guides to the outcome. 

Talking about the close Wisconsin Senate race between Republican incumbent Ron Johnson and Democrat Mandela Barnes, Marquette University political scientist Julia Azari said, “It looks like it will come down to whoever sneezes on Election Day.” 

The same can be said about the national results in what could be the first midterms decided by a stray achoo.

Walter Shapiro has covered the last 11 presidential campaigns. He is also a fellow at the Brennan Center for Justice at NYU and a lecturer in political science at Yale. Follow him on Twitter @MrWalterShapiro.