RCP's Interactive Tool: How to Explain the GOP Race

RCP's Interactive Tool: How to Explain the GOP Race
AP Photo/Mike Carlson
Story Stream
recent articles

Two weeks ago, RCP senior elections analyst Sean Trende and I attempted to explain how Donald Trump came out of nowhere to become the front-runner for the Republican nomination. One of our key points was that Trump had consolidated the blue-collar, immigration-hardliner wing of the Republican Party while his rivals divvied up the rest of the vote. And despite the insane rollercoaster of primaries and debates that have happened since our article came out, that factional friction still explains most of what’s going on in the Republican primary. Specifically, Trump’s ability to stay on track to get a majority of delegates (or come close to it), Texas Sen. Ted Cruz’s surge and Florida Sen. Marco Rubio’s decline are for the most part direct or indirect consequences of these persistent splits within the party.                                                      

In order to demonstrate those divides, I made (and have now updated) an interactive tool called the RCP Correlation Machine. It’s easy to use -- just select a candidate and a state or region from the drop-down menus and then press “go.” The Correlation Machine will then do a county-by-county analysis that shows which variables (both demographic and political) are the best and worst predictors of that candidate’s support in that area. My analysis is below, but if you play with it for just a minute or two before reading further, you’ll clearly see the divides within the party.

Before we get into the interpretation of these results (skip to the next paragraph if you’d rather stay out of the weeds on the math and data) I should explain correlation and briefly flag a few details about the data. Correlation is a measure of how well one thing predicts some other thing on a scale from negative one to one. Correlation will be negative when there’s a negative relationship (e.g. people who exercise more have a lower risk of heart disease), positive when there’s a positive relationship (e.g. students who study more get higher grades) and zero if there’s no relationship. If the correlation is close to one or negative one, the relationship is very strong. The data is mostly from official results. In Kansas and Minnesota some of the county vote shares were estimated, and we do not have results by borough for Alaska. We also did not have complete data for U.S. territories. Washington, D.C., is counted as one county, so no correlations could be calculated.

Trump is winning, but he’s not expanding his support. Right now, the billionaire businessman holds a solid lead in the delegate count and is on target to earn (or come very close to earning) the 1,237 delegates needed to win the nomination. Despite his successes, Trump isn’t gaining momentum -- the character and size of his coalition has been basically unchanged since voting began in early February.

In nearly every state, many of the same factors -- higher levels of unemployment, less education, lower income, and larger racial minority populations (often black) -- lead to higher vote shares for Trump. This makes sense. His positions on trade might appeal to those who have fared poorly in the modern economy -- blue-collar workers who might have been laid off or who haven’t seen wage increases in years. Trump’s less than racially sensitive rhetoric might also explain why he does well in counties with larger black populations. The decades-old “racial threat” hypothesis posits that whites who live in areas with large African-American populations are more susceptible to racialized appeals, and we could be seeing that play out here. Obviously not all of his support is explained by these factors alone, but the interactive shows that the correlations are strong.

Trump’s support isn’t expanding beyond this group, however. These same factors led to higher Trump vote shares before Super Tuesday (when he led, but not by as much), on Super Tuesday (when he had a very good night) and after Super Tuesday (during the Cruz surge). Trump’s share of the popular vote has hovered around 35 percent since voting began -- as have his national poll numbers. So Trump can lead a divided field, but voters outside his base still aren’t jumping on the bandwagon.                                                                                                          

Rubio failed (partially) because he and Kasich split the white-collar wing. In some ways, voters who like Rubio and Kasich are the opposite of Trump voters. Rubio and Kasich often do best where people are better-educated, have higher incomes and have only moved to the state recently. Their supporters are somewhat different ideologically -- Kasich attracts more moderate voters and Rubio attracts more conservative ones -- but they’re similar demographically.

Rubio’s failure to consolidate this bloc may have doomed his candidacy. Rubio started to fall behind on Super Tuesday, when he only won Minnesota and fell far short of the delegate count he would have needed be on track to clinch the nomination. If he had won even some of Kasich’s votes, he likely would have won a few more states and come close to Cruz’s delegate count. But instead he split those votes with the Ohio governor, narrowly lost some states, fell under the threshold for getting delegates in others and lost momentum heading into the subsequent contests (for more on this, see Dave Wasserman’s piece for FiveThirtyEight).

The Cruz surge is built on solidifying the very conservative lane. Right now, Cruz is surging. He scored a somewhat surprising win in Kansas and a very surprising win in Maine on March 5, and he followed it up with a win in Idaho last Tuesday. According to the Correlation Machine and other data, Cruz owes his surge to his conservative, Tea Party-style base.

The Correlation Machine didn’t find strong positive demographic predictors of the Cruz vote (large families and having children under 18 were positive but often relatively weak), but it did find strong, telling negative relationships. Cruz often does best where there are lower levels of support for Trump, Kasich or Rubio or where the demographics don’t favor any of those three. This suggests Cruz isn’t sharing some demographic group with another candidate.

Instead, he seems to be doing well with Tea Party voters. Exit polls and public polls often show the Texas senator doing well with voters who self-identify as very conservative and evangelicals who attend church weekly. These aren’t the angry populists of the Trump movement -- they’re more ideological, more activist and more religious.

This group is formidable, but they didn’t arrive early enough or in large enough numbers to put Cruz ahead of Trump. Trump won many of the Southern states where these voters are most numerous, so it will be tough for Cruz to put together a winning coalition without bringing another group into the fold.

What This Means Going Forward

If this election has taught political forecasters and analysts anything, it’s humility. So I won’t make any confident predictions of exactly how these demographic divides will (or won’t) shape the future of the race. But so far it’s been clear that Trump leads when the field is too divided. And with more winner-take-all and winner-take-most states coming up in the calendar, it’s easy to see how a splintered party could end up with a Trump nomination or contested convention.

David Byler is an elections analyst for RealClearPolitics. He can be reached at dbyler@realclearpolitics.com. Follow him on Twitter @davidbylerRCP.

Show commentsHide Comments