There is a general theory of how politics works in the United States right now that goes something like this:

Democrats and Republicans advance policy agendas that are covered in a straightforward manner by the media and broadly digested by voters. Based on their views of these agendas, most voters then align themselves with one of the two parties. But there is also a group of voters with views that fall roughly between the parties and therefore are not aligned with either. Those voters “swing” between the parties, usually choosing the one whose agenda is closest to the ideological middle. Thus, elections are largely decided by whichever party wins the most of these centrist voters. Voters will also swing decidedly against a party if a president from that party screws up something big.

This is basically all wrong, or at least badly overstated. And worse, belief in this myth warps our understanding of what’s really going on. We can’t see our system for what it is if our lens for looking at it is distorted. Here are seven better explanations of U.S. politics today:

1. Americans’ views on most issues aren’t deeply held — and can be highly influenced by their party, the media or both.

In 2019, Democrats were wary of moving to impeach then-President Donald Trump, in part because polls suggested doing so would be unpopular. But once congressional Democrats announced they were starting the impeachment process, support increased substantially, in part because it surged among Democratic-leaning voters. This was not unusual — there is a lot of evidence that Americans’ views on many political issues aren’t set in stone and will shift in the direction pointed toward by high-profile figures in their party or in media they trust.

This dynamic makes me skeptical of polling on issues, such as what percentage of Americans say they support President Biden’s infrastructure plan. I think polls do give you a generic sense of what kinds of ideas are very popular or very unpopular to people at first glance. But let’s say that Biden, House Speaker Nancy Pelosi (D-Calif.) and former president Barack Obama emphatically endorse a fairly controversial idea, such as adding justices to the Supreme Court, and mainstream media outlets such as the New York Times, The Post and CNN that are heavily consumed by Democratic-leaning voters cover those statements in a non-negative manner. I would expect support to shoot up, particularly among Democratic-leaning voters.

Similarly, though polls suggest that Biden’s infrastructure plan polls pretty well even with Republicans, I think congressional Republicans could easily drive down support among GOP voters if they tried hard. An example: Polls nearly always show high support for expanding background checks on gun sales. But in 2016, a Nevada ballot initiative on background checks barely passed, likely because prominent state Republicans opposed it and persuaded GOP voters to do the same.

2. Besides, views on issues don’t predict election outcomes.

Follow Perry Bacon Jr.‘s opinionsFollowAdd

The other problem with issue polling is that the results often aren’t correlated with election outcomes. What is generally implied in political reporting is that a party takes popular or unpopular stands; voters learn the details of these positions through media coverage that presents the parties’ stances clearly and accurately; and people, particularly swing voters, vote based on those stances.

But it’s easy to see how this model doesn’t reflect real life. Most people don’t know much about policy. Few outlets cover policy in detail, and some of them, such as Fox News, go out of their way to present policy issues in false or misleading ways. Plus, no one is required to vote based on policy — and people often don’t. For instance, there is evidence that Americans view individual candidates who are women or Black as more liberal than their actual platforms, because they are part of larger groups (women and people of color) that are left-leaning.

In our current era, Democrats obsess about things that have more than 60 percent support in the polls, such as health care and bipartisanship. Republicans do things that polls suggest have less than 40 percent support, such as trying to repeal Obamacare and not being bipartisan. But our elections are close to 50-50 anyway — which suggests that these strategies are about equally effective and that doing what polls well doesn’t boost Democrats much.

3. Governing performance often doesn’t matter.

President George W. Bush’s mishandling of Hurricane Katrina and the Iraq War caused him to sink in the polls, helping Democrats sweep the 2006 and 2008 elections. But partisan preferences are much more locked in now than they were even 13 years ago. Trump’s approval ratings didn’t go up much even as the United States had historically low unemployment numbers before the pandemic, nor did they go down much during his terrible handling of the outbreak of covid-19. Biden’s generally competent handling of the pandemic has not increased his support. Instead, Biden’s approval ratings are basically just the inverse of Trump’s.

So while Biden’s team crows about getting vaccines in arms and money in people’s pockets, there is little evidence that being good on policy is a huge electoral boon, particularly since in today’s media environment, Americans may not hear accurate information about these policy successes.

4. Swing voters aren’t all centrists.

One explanation for the limits of issue polling and governing performance is the false assumption that the electorate includes a mass of ideological moderates who are drawn to the party that governs best or in the most centrist manner. But many of the voters we describe with terms such as undecided, swing, moderate, centrist and independent either don’t have a firm set of issue beliefs or hold a jumble of views that don’t line up with either party. They might support building a wall on the U.S.-Mexico border and also legalizing marijuana, for example. Relatively few voters are true centrists, with views roughly equidistant from the two parties. This dynamic explains how a sizable bloc of voters supported Obama in 2012 but Trump four years later.

5. And one of the biggest swings in elections is who turns out.

Millions of Americans who voted in 2016 didn’t vote in 2018, and there was a small but still sizable bloc that voted in 2018 but not 2016. Estimates suggest that nearly 40 million voters (25 percent of the electorate) in 2020 were people who didn’t vote in 2016. Michael Podhorzer, a senior adviser at the AFL-CIO, argues that midterm elections in particular are determined by turnout. The president’s party often loses in midterms, according to Podhorzer, not because of some backlash against the president from swing voters, but rather because voters from the party out of power are more motivated to vote.

In 2020, Trump did better than the polls projected, and one explanation is that polling missed the millions who didn’t vote in 2016 but backed Trump last year. In particular, much of Trump’s gains with Latino voters came from people who didn’t vote in 2016. So although people who vote consistently but change parties really matter, so do those who swing into and out of voting.

6. Plus, wooing “swing” voters isn’t in tension with winning “base” voters.

You may be thinking — but isn’t the victory of Biden, who pitched himself as a moderate (even if the platform he ran on was among the most liberal of any nominee in recent history), a validation of centrism? Yes and no. Biden did win more votes from independents than Hillary Clinton did, but he also won more votes among millennials and African Americans. Much political coverage assumes that appealing to a middle-age White voter who swings between the two parties requires a different campaign message than the message that would appeal to a Black person who leans Democratic but doesn’t vote in every election. But “turnout” and “persuasion” are not competing approaches. An effective campaign is likely to swing consistent voters to its side while also swinging sporadic voters into casting ballots.

7. In the end, politics is unpredictable.

None of this means that being way to the left or to the right, ignoring the true swing voters or governing badly will help a politician. But it suggests that the tactics that we think are optimal (governing well, not being too extreme) have somewhat limited benefits. The reality is that politics is complex and somewhat unpredictable — with a lot of voters who are regularly changing or just forming views on issues and also changing who they vote for or whether they vote in the first place. “Joe Biden won because he is moderate, so we can win as long as we stay moderate” may be a comforting view for Democrats. But it’s not a reality-based one.