From The 100 Million Project: The Untold Story of American Non-Voters.

Quick, who won the popular vote in the 2016 presidential election?

The conventional answer is Hillary Clinton, but neither Clinton nor Donald Trump got nearly as many votes as the number of eligible voters who did not cast a ballot at all. As a new Knight Foundation report reminds us, if we looked at all eligible US voters, the result would have looked like this:

Did Not Vote41.3 percent
Hillary Clinton28.5 percent
Donald Trump27.3 percent
Other2.9 percent

Turnout, as political candidates know, matters. This is true even if political pundits seem to routinely forget it. And, if you don’t think pundits regularly forget this, I encourage you to start counting how many times pundits on cable broadcasts and in the press assert, contrary to all evidence, that winning the middle-of-the-road voter is key to victory.

And it’s not just the turnout rate that matters. It is who turns out that matters the most. In 2012, for instance, Barack Obama sailed to reelection in what was supposed to be a close race, largely due to a record Black turnout. I don’t know what the average NPQ reader recalls, but I will confess that I had forgotten that the pre-election Gallup polls showed Mitt Romney leading Obama (which, by the way, makes pre-election polls wrong in the last two consecutive presidential elections—a warning if we needed one not to place too much confidence in polling instruments).

By contrast, the turnout rate of Blacks fell seven percentage points (from 66.6 percent to 59.6 percent) between the 2012 and 2016 elections. This had a major impact on electoral outcomes. All told, in the 2016 elections, an estimated 4.4 million voters (over a third of whom were Black) who had voted for Barack Obama in the 2012 election stayed home, easily enough to have shifted which candidate prevailed in key swing states such as Michigan, Wisconsin, and Pennsylvania.

The Knight Foundation report is titled The 100 Million Project: The Untold Story of American Non-Voters. Whether the story is “untold” is questionable, but the topic is clearly important. Knight did create a very large polling data set. As the report details, this includes:

  1. A national voter sample of 4,002 consistent nonvoters (over the past decade).
  2. A larger “swing state” sample of 8,015 consistent nonvoters in Arizona, Florida, Georgia, Michigan, Minnesota, Nevada, New Hampshire, Pennsylvania, Virginia, and Wisconsin.
  3. A national sample of 1,002 voters (people who voted in at least three of the last six presidential and midterm elections).
  4. A young voter sample of 1,035 voters. Because this sample focused on voters 18 to 24 (many of whom, of course, were ineligible to vote in 2016), this sampling includes both voters and nonvoters.

Why is US turnout lower than in most countries? Two years ago, Drew DeSilver, a senior writer at the Pew Research Center, looked at the dynamics of turnout and highlighted the fact that as a percentage of registered voters, US turnout is actually pretty high. In 2016, for instance, 86.6 percent of voters who were registered voters did vote. But overall turnout was less than 60 percent.

Except for Luxembourg, no country exceeds the US in the gap between the number of eligible voters and registered voters. As DeSilver writes, “Registered voters represent a much smaller share of potential voters in the US…only about 64 percent of the US voting-age population (and 70 percent of voting-age citizens) was registered in 2016…compared with 91 percent in Canada (2015) and the United Kingdom (2017), 96 percent in Sweden (2014), and 99 percent in Slovakia (2016).

What causes the gap? Well, voter suppression, most often subtle (making it hard to register), but also overt, is a large part of the explanation. As NPQ noted earlier this year, it is not too mysterious how to increase voter participation. As Australia demonstrates, if you automatically register people to vote and make voting compulsory (even if the penalties for not voting are minuscule), then voting rates soar.

As Australia’s recent electoral history also demonstrates, increased turnout does not necessarily lead to more “left” or “liberal” (in US parlance) voting outcomes. The Knight Foundation data also show this. For instance, Knight found that 40 percent of voters had a favorable opinion of Trump, as did 40 percent of nonvoters.

For better or worse, the Knight survey puts a lot of energy into election polling, but this mainly confirms the general finding that nonvoters and voters have similar preferences. Nationally, surveyed nonvoters favored a hypothetical Democrat over Trump by an underwhelming 33 percent to 30 percent margin, with a 1.55 percent margin of error. “Swing state” surveys, with a 3.5 percent margin of error, found nonvoters leaning toward an unnamed Democratic candidate in Georgia, Minnesota, and Wisconsin; evenly split in Nevada and Minnesota; and favoring Trump in Arizona, Florida, New Hampshire, Pennsylvania, and Virginia. In many of these states, the margin of error exceeded the size of either party’s lead.

Even though candidate preferences may be similar for voters and nonvoters, voter suppression can still be disturbingly effective, especially if there is a racially disparate impact in who is denied the ballot. Prior to Georgia’s 2018 elections, for example, 70 percent of suspended voter registrations were of Black voters, even though only 32 percent of eligible voters were Black. Again, it is not just the total turnout that matters, but who turns out that affects outcomes.

Not surprisingly, while nonvoters do not differ markedly from voters in their views, people aged 18 to 24 did. As the report indicates, “With regard to policy, young people elevate a different set of priorities than voters or non-voters, with gun control, the environment, and race relations more prominent among their top issues.… If they were all to vote in the 2020 presidential election, young people are twice as likely to support the Democratic candidate rather than President Trump.”

But even when comparing nonvoters and voters, superficial parity may obscure key differences. Past studies have shown that nonvoters tend to be more left-of-center on economic issues (e.g., tax policy, social spending) and more right-of-center on so-called social issues (e.g., immigration, abortion, gun laws). Unfortunately, the Knight study does not ask about such core economic issues as minimum wage and taxation, where these divides might appear.

As Sean McElwee, executive director of Data for Progress, observed in the Atlantic years ago, “measures that examine a one-dimensional left-right axis obscure these divides.” This is a problem where, unfortunately, the Knight study is largely guilty as charged.

The only quasi-economic issue that the Knight study asks about is the Affordable Care Act, a problematic issue to focus on given that nonvoters—which, Knight’s data show, are poorer than voters—are more likely to still be uninsured and/or underinsured despite the legislation. So, we shouldn’t be too surprised that nonvoters are somewhat less keen on the policy (although majorities of both voters and nonvoters favor keeping the law in place).

The Knight report also divides nonvoters into six profiles, like a corporate marketing study might do. These categories are “established progressives” (21 percent of nonvoters), traditional conservatives (17 percent of nonvoters), modern moderates (20 percent of nonvoters), indifferent average (17 percent of nonvoters), unattached apoliticals (17 percent of nonvoters), and underemployed unsures (eight percent of nonvoters).

The first three categories, broadly speaking, consist of people who are “plugged in” but still don’t vote. The latter three, by contrast, consist of people who the study labels “disconnected,” but in any event are evidently alienated from the US political system. It is not clear how seriously we should take these segments. But they do speak to the diversity that exists within the broad “nonvoting” category of 100 million people.

Perhaps more meaningful are the demographic data, some highlights of which are listed below:

Percent of VotersPercent of Nonvoters
People of color28%35%
Voters age 25 to 4030%40%
Earn over $100,000 a year29%13%
Earn less than $50,000 a year26%44%

These data show that there is at least potential room for a candidate who can mobilize people who are low-income, of color, and young to change the electoral map. Being successful at this, of course, remains a challenge, but as Obama’s campaigns from 2008 and 2012 demonstrate, not necessarily an insurmountable one.

Understandably, given the topic, Knight focuses on making direct comparisons of voter and nonvoter preferences, with an eye to uncovering differences. But for those concerned about the health of US democracy, some of the data from voters is disturbing.

For example, only a third of voters and nonvoters think the country is “on the right track” while 61 percent of voters and 54 percent of nonvoters think it is on the wrong track (the rest indicating “not sure”). Overwhelming numbers agree that “moneyed special interests have greater influence over elections than voters” (82 percent of voters and 72 percent of nonvoters strongly or somewhat agree), compared to 11 percent of voters and 13 percent of nonvoters who disagree. Less than two-in-five voters and nonvoters (35 percent of voters, 39 percent of nonvoters), agree or somewhat agree that “The federal government works to improve the lives of people like me” (53 percent of voters and 49 percent of nonvoters disagree).

In addition to the polling, five focus groups provided some qualitative data. Unfortunately, this data is mainly used to generate pull-quotes rather than frame analysis, but some of the quotes highlight a deeper unease:

  • “The times that I’ve spent to get a little bit more educated, all the options suck.”—Female non-voter, Milwaukee
  • “I hate the thought of voting for the lesser of two evils. That just turns my stomach.”—Female non-voter, Las Vegas
  • “Voting is very hard to prioritize. If I need to feed my kids or if I need to vote, I’m feeding my kids.”—Male non-voter, Philadelphia
  • “I don’t have faith in the electoral process or faith in the current government. Most of the representatives don’t look like me, don’t sound like me, don’t come from where I come from, and don’t represent who I am.”—Female non-voter, Philadelphia

It is not news that faith in government in the US has declined for decades. Back in the 1960s, the percentage of Americans saying they trusted “the government in Washington always or most of the time” remained consistently far above 50 percent. It dropped to 36 percent after Watergate—and, except right after the 9-11 attacks, has never been above 50 percent since. As of March 2019, that number stood at 17 percent.

Knight’s report does not offer much in the ways of solutions to this conundrum. But it does have the merit of naming the problem. The report authors conclude, “We believe that informed and engaged communities are the foundation of a healthy democracy. Voter participation is a fundamental form of democratic engagement, yet the United States lags most developed democracies in terms of eligible voter turnout. A deeper understanding of what non-voters think and how they behave is critical to understanding the present and future of our democracy.”