For all the talk of Donald Trump’s demise his standing in national and state polls has never been stronger. But it has rarely been weaker either. Trump’s support has largely stayed static as support as shifted between lower tier candidates.
You would not notice this if you looked at recent polls including one in particular, the latest CNN/ORC Opinion survey. The survey is an outlier among outliers, showing Trump leading the GOP field nationally with 36 percent and Cruz next at 16 percent. Few doubt Trump is ahead in the GOP race. But by 20 percent?
Contrast the CNN survey with a Quinnipiac survey that found him with a more modest 27 percent of the GOP vote compared to Rubio at 17 percent and Carson and Cruz at 16 percent.
Then we come to the methodological issues that have plagued surveys since last cycle. Last year, polling dramatically underestimated GOP support in Senate and Gubernatorial races. More recently, in Kentucky, both local and national polling was well off the mark in Kentucky’s gubernatorial contest.
These issues are certainly not just an American political issue. Beyond last year’s midterms the polling also missed Israel’s right-wing Likud Party besting their left wing opposition. It also substantially underestimated Britain’s Conservative Party support. This year, the polling was dramatically off in Canada’s provincial and national elections.
But nomination contests offer a different set of issues, particularly national contests. Specifically, the sample and sample size one uses in surveys. Due to the unusual combination of primaries and Caucuses involved in Presidential nomination battles actual turnout can vary widely. So while turnout in the Iowa Caucuses might be 10 to 12 percent of all registered Republicans in the state turnout in the New Hampshire primary might want a much higher percentage of Republicans turnout out.
The way many pollsters sample voters, random digit dialing (RDD), makes many question whether those polled will actually turn out. Secondly, the weak nature of sampling filters at this time makes it likely some voters saying they are Republicans really are not or have limited interest in the contest.
Polls have already played an outsized role in the GOP contest. Every major news network to broadcast a GOP debate has designed its criteria around polling results. Many of the leading contenders in the GOP field, Trump, Cruz, Rubio and Carson have been able to stay in the debate largely because they have been able to broadcast their views to millions of people at one time (over and over).
By far the biggest issue with all polling (and Trump’s lead) is over the way polls assemble their sample of voters. Assuming every Republican candidate has filed to get on the ballot, each can earn delegates in 50 states, the federal district and five overseas territories and each of those has its own rules about how those delegates are awarded. Some have primaries that are closed or open to registrants of other parties, caucuses and party conventions in which rank-and-file Republicans don’t even get a vote, etc.
It’s thus not surprising that many analysis and Republican strategists feel the nomination battle is partly being shaped by voters who will not participate in the process and by inaccurate polling.
Republican pollster B.J. Martino took apart the CNN survey thusly and also described what is the issue with many national polls showing Trump with a wide lead, “They have a sample of 1,020 adults — and 445 of those they say constitute the Republican primary universe. Basically, their poll is saying 43.6 percent of all Americans adults are voting in a Republican primary nationwide. When you go back to 2012, it’s 12.2 percent.”
Put simply, pollsters are including in the 2016 GOP primary samples a group 3.5 times bigger than 2012. Unless the GOP is in for a wave election that assumption is unrealistic.
Trump does have several things going for him this late in the game. First, the typical accuracy of polling has been off of late. Perhaps these voters will actually turn out for Trump after-all. Secondly, national polls tend to overestimate the number of moderate and liberal voters who participate early in the process (Blue-state Republican primaries are usually later in the process when fewer candidates are left running). The actual electorate early in the process is pretty conservative.
But Trump’s support is pretty consistent across most groups. In the Quinnipiac and CNN survey Trump found pluralities of support among men, women, college educated and non-college educated voters, conservatives and moderates.
Trump’s support can largely be attributed more to his attitude than views. A recent analysis by the WSJ found in polling from July to October that Trump had support largely from voters who were drawn to his view as a “strong leader” and less so on particular issues. While predominately blue-collar, those open to voting for Trump split pretty evenly along class and ideological lines.
By not being a base candidate Trump can take unconventional positions on the usual issues and be fine. In the recent Quinnipiac survey, 46 percent of Trump supporters said their minds were made up. If true it indicates that Trump voters are committed to their candidate and turnout in the primaries and caucuses could easily exceed 2012.
Of course, that is a big assumption. Assuming Trump maintains his leads in early states like Iowa and South Carolina is risky. His current support among evangelicals may not hold by February and Iowa and South Carolina have large contingents of the group. Also assuming Trump can in the SEC Primary and in Florida is a big “if.” There, evangelicals often cut into the blue-collar wing of the party and it does not make as easy a path for victory for Trump.
Like-wise, by the time the SEC Primary is upon us (after the Nevada Caucus), the GOP field will be whittled down. If it is a two-person or three-person by then the polling will be vastly different as will the size of Trump’s lead (unless he builds on his support which is unlikely).
Trump has upended the conventional logic this cycle and he does have the establishment worrying he could be the GOP nominee. But it is still far from a sure thing he will be, especially when the polls showing him ahead have such a shoddy track record of late and are assuming a massive GOP electorate showing up for next year’s nomination contest.