What happened? The common story is that McGovern was too left-wing and handicapped by having to replace his running mate after he was revealed to be suffering from depression (a big deal at the time). He also suffered from not having organized labor's backing. AFL-CIO leader George Meany declined to endorse McGovern, calling him "an apologist for the Communist world."
But as Andrew Gelman points out, Nixon also had the benefit of a strong economy. He points to this chart from the political scientist Douglas Hibbs:
But take a look at the race below 1972 on that chart. In 1968, the economy was growing about as fast, and yet the incumbent party – represented by Hubert Humphrey — lost. What gives? Hibbs has an answer. His model, called "Bread and Peace," takes into account both economic growth and the rate of military casualties abroad. The latter variable doesn't usually come into play, but in 1968, it did. Casualties were mounting in Vietnam, enough to blunt the economic factors.
So what changed in 1972? Wasn't the United States still fighting in Vietnam? Yes, but casualties had dropped dramatically. In 1968, over 16,000 American servicemen were killed in action. By 1972, that number had dropped to less than 1,000:
This suggests McGovern wasn't just doomed by the economy. If Nixon had failed to reduce casualties in Vietnam, the same factors that lead to Humphrey's defeat could have put McGovern over the top. But the war was winding down in terms of U.S. casualties, even as it expanded into Cambodia and Laos, subjecting many more Southeast Asians to the bloodshed.
McGovern didn't just have to beat a president who oversaw a booming economy. He had to beat a president who oversaw a booming economy and had, as far as American casualties were concerned, drawn down a war. According to Hibbs's model, that doomed him doubly.
That being said, McGovern still underperformed most models. Ray Fair's model, which relies mainly on economic factors, missed in 1972 by 3.3 points, one of its biggest errors ever, while Robert Erikson, Michael MacKuen and James Stimson's "Macro Polity" model, which looks at where the electorate sits in terms of partisanship and ideological beliefs, predicted that McGovern would do about two points better than he did. Hibbs's model also underestimated Nixon's vote share by about two points.
But McGovern's loss has been robbed of its context over the years. The fundamentals were extremely bad for a challenger. He was in for a drubbing even if he had done everything right.