The Washington PostDemocracy Dies in Darkness

Political science (and logic) suggest that Donald Trump’s infrequent-voter plan is risky

GRAND RAPIDS, MI - DECEMBER 21: Guests wait to greet Republican presidential candidate Donald Trump at a campaign rally on December 21, 2015 in Grand Rapids, Michigan. (Photo by Scott Olson/Getty Images)

Over the weekend, The Post's Jenna Johnson took a look at the disconnect between Donald Trump's boisterous support at his campaign events and the robustness of his positioning for the Iowa caucuses. Lots of people show up for Trump's stream-of-consciousness performances. But lots of them are also a bit murky on whether or not they'll actually cast a vote.

Voting of any kind involves a commitment of time and energy by the voter, as we all know very well, but Iowa's caucuses are another thing entirely. The locations change regularly and involve speeches before votes are cast. It takes a while. Trump's base of support includes a lot of people who aren't used to that and haven't caucused in the past, as a recent Monmouth University poll showed. The question, then, is: Will they actually come out on February 1?

Trump's campaign has a big index of supporters from its events, but hasn't bought the voter data that would tell it how to prioritize turnout, Johnson reports. That's problematic,too.

Imagine three voters: Aaron Five-of-five, Bill Three-of-five and Cal Zero-of-five. Aaron has participated in five of the last five Iowa caucuses, Bill, three and Cal, none. Let's say that all three went to a Trump rally and all three said they planned to caucus for the candidate. In the days leading up to the caucus itself, who should Trump spend his time trying to hustle to the caucuses, assuming he has a finite ability to contact everyone?

The smart answer is Bill, the guy who may or may not come out and make his voice heard. There's no need to bother Aaron; he's going anyway. And Cal shouldn't be a priority either, because the odds that this is the time he finally gets off his duff are pretty small. Trump's folks should be bugging the whole Three-of-five family for the last week of January -- but, per Johnson, Trump has no idea who those people are. So he'll waste a lot of time and money telling Aaron and Cal to go vote when those two are almost certainly going to continue their past practice.

This is a problem for Trump well beyond Iowa. Trump's base of support is heavily skewed toward less educated Republicans, a group that traditionally overlaps with those with lower incomes. Lower income voters are almost always also less likely to turn out to vote, as Census data makes clear.

Why? Lots of reasons. First, the required investment of time and energy. Lower income people are more likely to work unusual hours, making it potentially tougher to get to the polls. They're also more likely to move around regularly, not owning a home, meaning they need to constantly re-register to vote. Income tends to overlap with age and race, as you might expect, which helps explain the surge in turnout among lower income groups in 2008 and 2012 -- when Barack Obama energized a lot of young and black voters.

That's sort of what Trump is hoping will happen: He'll energize these folks and they'll get to the polls. Obama, though, put a massive investment in a field operation that would turn them out. Trump, so far, hasn't -- and time is running out.

It's also not safe to assume that people who say they're committed to coming out to vote for the first time -- proclamations that Trump loves to retweet on Twitter -- will actually do so. Echelon Insight's Patrick Ruffini pointed to a 2011 study from Harvard's Kennedy School of Government looking at self-reported intention to vote. "Accuracy [of the voter's prediction of his behavior] was unrelated to time before the election, but was significantly affected by consistency with voting in recent similar elections. Specifically, accuracy was greatest when predictions were consistent with past behavior." In other words: People who say they are going to vote are a lot more likely to actually do so if they have voted in the past. Bill Three-of-five is more believable in this regard than Cal Zero-of-five.

Larry Sabato of the University of Virginia thinks the Monmouth poll cited above and a poll from Winthrop University in South Carolina are a better reflection of the outcomes in those states because they constrain the pool of people asked for their opinions to those with a record of having voted in the past. In those surveys, Trump gets 19 and 24 percent support respectively, compared to the 27.5 and 33.7 percent Trump has in Real Clear Politics's current polling averages.

Elections aren't magic. People do get jazzed about candidates and sometimes vote for the first time out of sheer enthusiasm. But voters are usually much more predictable than that. Donald Trump is running an unorthodox campaign and counting on an unorthodox outcome. As Jenna Johnson reported, though, even the people who show up to Trump rallies are often lukewarm on being part of that magical movement.