Andy Larsen: How do caucus goers differ statistically from primary voters in Utah? Investigating the data.

Last week, the Utah Republican Party announced its decision to not hold a primary next year. Instead, it’s a return to old-school, in-person caucuses to decide its candidate for president of the United States of America.

My colleague Robert Gehrke already wrote about the subject. His story included quotes from Gov. Spencer Cox about how the last time the GOP ran a caucus in 2016, those delegates selected “did not represent the Republican Party at all.”

My question: is that really true? I think we all agree that caucuses are more inconvenient than their primary equivalents, especially when by-mail voting is the standard in traditional primaries. But how much does that actually change the electorate? Can we quantify those changes?

Turnout down in caucuses

There’s no denying it: Caucuses are terrible for turnout when compared with standard primaries. As Gehrke noted, 177,000 Republicans participated in the 2016 Utah caucuses. When the party used a traditional primary in 2020, about 345,000 Republicans participated.

This is a nationwide trend. D.C. think tank Third Way put together this helpful comparison of every state’s primary and caucus turnouts in 2016 when compared to 2020. Note that they compared Democratic primaries and caucuses instead of Republican ones, because there was a new presidential candidate to elect in both years. (In 2020, the Republican nominee was obviously going to be the incumbent, Donald Trump.)

Full stop, every caucus state underperformed every primary state. It’s actually stunning. Even Iowa — which runs a caucus in the first election event of the primary season and therefore has outsized importance for the rest of the country — couldn’t garner more folks than a run-of-the-mill primary in noted non-Democratic hotbed Tennessee.

In 2020, multiple states didn’t fund presidential primaries for their parties, so five states’ Democratic parties held party-run primaries, where they used various party-supported methods to get ballots from their voters. Those didn’t work very well in getting turnout, and neither did caucuses.

Note Utah’s jump from 4.2% turnout in 2016 to 10.2% turnout in 2020 on the Democratic side. Utah was among one of seven states who switched from caucuses to primaries in between the two years, along with Colorado, Idaho, Maine, Minnesota, Nebraska and Washington. A combined 746,000 people participated in their 2016 caucuses, according to Third Way; a combined 3,962,000 people participated in their primaries come 2020. Again, a staggering difference.

A changing electorate

So clearly, fewer people show up overall for caucuses than primaries. But that doesn’t necessarily mean the results are any different. If three people vote for Trump and two for Ron DeSantis in a caucus, but six people vote for Trump and four people vote for DeSantis in a primary, it really doesn’t matter that much — the margins are 60% to 40% in Trump’s favor either way.

So in addition, we need to look at the voting body’s characteristics change in caucuses than primaries.

This turns out to be a harder question to answer than I anticipated. I expected mountains of public research on the topic, but there really wasn’t much.

The problem comes down to how we understand who votes in an election. On the ballot, there’s no question about gender, race, income etc., and rightfully so. That means in order to learn who shows up to vote, we rely on high sample-size exit polling to understand who turned out. In college, I participated in the Utah Colleges Exit Poll in the 2012 election — which meant I stood outside of a polling place and asked people who had just left the voting center about who they were and how they voted. Overall, we polled 13,725 voting Utahns across the state that day. Pretty great!

But exit polls aren’t super common for primaries, and are quite uncommon for caucuses. So without that, we have limited information about who voted and why.

One study, though, gave me the idea I used in this column. Costas Panagopoulos, a professor of political science from Northeastern University, wrote a piece called “Are Caucuses Bad For Democracy?” in 2010. In order to study the turnout, and in lieu of exit polling data, he used data from the Cooperative Election Study (CES), a well-regarded poll that interviews over 50,000 Americans every election cycle.

One question the pollsters ask is whether a respondent voted in the primary or caucus, so we can just look at those who said yes. One reason that we know it’s a good poll is that it verifies a respondent’s yes or no answer through public voting records.

Unlike Panagopoulos, I want to zoom in on Utah. So I looked at 631,000-odd total participants, got just the data from the 2008, 2012, 2016 and 2020 presidential election years, and compared three different groups:

• All Utahns who responded to the CES

• Utahns who voted in a primary

• Utahns who voted in a caucus

Here’s what I found:

In many demographic metrics, it’s a sliding scale: people who tend to vote in primaries at all are already a little different than the overall population. People who participate in caucuses tend to be even more different.

So, for example, 49.9% of responding Utahns were male, but of those who participated in primaries, nearly 56% were male. And in caucuses, just over 60% of participants were men. The same is true of race — caucus goers tended to be a little more white than the rest.

They also tended to be older — the average respondent to the poll was 44, while the average caucus goer was 53. But other related metrics change even more than you can explain with age alone. Caucuses had nearly double the number of retired people than the overall population would indicate. Those who had kids under 18 were significantly less likely to attend a caucus.

Age is a part of this story, but so too is time; those who are retired and without children just have more time to attend a caucus night than those who don’t. Caucus goers were also more likely to be single than the overall population — though politics nerds might consider a caucus night a great date night, most folks probably wouldn’t.

I also thought the health insurance aspect was interesting. Nearly everyone who attends on caucus night has health insurance, at much higher rates than the rest of the population. I can’t blame anyone who chooses not to attend a crowded caucus event in which disease is more likely to spread if they don’t have the insurance to deal with that potential sickness. This also may result in candidates who care less about universal health insurance than the population does overall.

It’s worth noting, though, that there were many metrics that didn’t show big differences between those who attend caucuses and those who voted in traditional primaries. Take income: about 20.7% of people at caucuses made between $10K and $40K per year; that number was about 20.8% for primaries. The difference in education wasn’t huge, either: 57.1% of primary voters had a college degree, compared to 58.6% of caucus goers.

More renters voted in caucuses (20% of caucus voters) than primaries (17.2%). That being said, renters made up a far smaller part of both groups than they do in the state overall (32.7%).

In Utah, religion often plays a political role. Both primary voters and caucus goers were significantly more likely to attend church every week (or more) than those who didn’t vote in either. Both were more likely to be members of The Church of Jesus Christ of Latter-day Saints, but in this category, primary voters (54.9%) were slightly more often adherents to that faith than caucus voters (53.6%) or the respondents overall (50.6%).

There are a couple of ways to take this data. Panagopoulos was among those who saw only a few percentage points worth of difference in gender, race and education between caucus and primary voters and felt that it showed that caucuses weren’t, in his words, “especially bad for democracy.” Caucuses are generally cheaper than their primary counterparts.

On the other hand, any difference between voting groups can be significant in razor-thin elections, and in some categories, the difference between those who vote in caucuses and primaries can be quite large. For example, Utahns overall persistently say that improving K-12 education is one of their most important issues. But if less than 20% of caucus goers have kids in that system, are the delegates or the candidates they elect likely to prioritize it? One Hinckley Institute study from way back in 2011 said no.

I tend to fall on that side of things. To me, it’s hard to justify holding caucuses instead of primaries given the turnout issues discovered from the data. Government probably needs more votes from all cross sections of society, rather than just those willing to spend a night with their neighbors talking politics.

Andy Larsen is a data columnist for The Salt Lake Tribune. You can reach him at alarsen@sltrib.com

Editor’s note • This story is available to Salt Lake Tribune subscribers only. Thank you for supporting local journalism.