Last week’s Crystal Ball repeated its refrain that Donald Trump is very unlikely to get the Republican presidential nomination. However, it issued the caveat that “[i]t would be easier to make our argument if we could explain precisely how and by whom the real estate tycoon will be dethroned.”
Who can and will defeat Trump? The answer is obvious, but also not obvious because the question seems to beg the name of another candidate. The “who” that will defeat Trump is not another candidate but is most likely to be the Republican voters who actually turn out in Iowa, New Hampshire, South Carolina, and the other contests.
About 75%-80% of the Republican primary voters in the early states are not likely to vote for Trump. How can that be? Aren’t the RealClearPolitics and other polling averages now showing Trump getting 25% to 35% or more in those states?
The Crystal Ball hinted at the answer this question in the same analysis when it accurately pointed out that, “[t]he billionaire’s appeal is very disproportionately tilted to the blue-collar half of the electorate…” Such voters are substantially less likely to actually turn out than are upscale voters, especially in caucuses, but also in primaries.
Evidence for this is clearly shown in six polls reported recently — four in Iowa (Monmouth, Des Moines Register/Bloomberg/Selzer, Quinnipiac, and CNN/ORC) and two in South Carolina (Winthrop and Fox News). The Monmouth and Winthrop polls use sampling methods that are about as good as a pollster can use at this point in the process — i.e., starting with a list of registered voters from state records and randomly sampling from only those who have voted in previous caucuses or primaries. To take into account both new registrants and also voters who were already registered but did turn out in previous nomination elections, the pollsters supplement their samples of those who did with additional voters who now express strong interest in this year’s nomination battle. Selzer starts with the active registered voter list and screens aggressively from there to get a sample of voters actually likely to brave Iowa’s winter cold for an entire evening on Feb. 1.
By contrast, the CNN, Quinnipiac, and Fox News polls assume almost all the registered Republican voters will vote in the Iowa caucuses and South Carolina primary, a deeply dubious assumption. Why?
In 2008 and 2012, Republican turnout in the Iowa caucuses was about 120,000 voters. Because of this year’s political excitement and interest turnout is likely to increase — perhaps to 150,000 or higher. Monmouth’s assumption is 140,000. By contrast, the CNN poll implicitly assumes Iowa Republican turnout of 500,000 or more — not plausible. (The Iowa Secretary of State’s Oct. 1, 2015 report counts 609,853 active registered Republicans in the state). Most of them are unlikely to turn out on a cold Monday at the start of February. The situation in South Carolina is similar, with the caveat that it is easier to vote in a standard primary than in a lengthy caucus process. But, again, the Fox poll estimate appears way too high while the Winthrop sample seems quite realistic.
In Iowa, the CNN estimate has Trump getting 33% of the vote while Monmouth has him at 19% — 14 percentage points, or 42%, lower. The DMR/Bloomberg/Selzer poll has Trump at 21%, 10 points behind Ted Cruz.
In South Carolina, Fox has Trump at 35%, while Winthrop has him at 24% — 10 points or 32% lower.
Trump’s true level of support is much more likely to be near the Monmouth, Selzer, and Winthrop estimates. That puts him in the 20%-25% range, not 30%-35%.
Trump has alienated many other Republican candidates and their followers. As second and third tier Republican candidates drop out after poor performances in Iowa, New Hampshire, and South Carolina, it is unlikely that Trump will pick up many of their voters. Rather, those votes will mostly go to other top tier Republicans, both outsiders (Ted Cruz) and insiders (Marco Rubio and Jeb Bush).
Trump’s “ceiling” appears to be hovering at 25% to 30%. This is too low to do well in the winner-take-all Republican contests starting on March 15 in states like Florida and Ohio. There are several other winner-take-all contests throughout the rest of the primary season, requiring majorities or big pluralities to win significant numbers of delegates.
It is nearly impossible to say now who will get the Republican nomination, but it is unlikely to be Trump. It will be the voters who end his quest for the presidency based on their evaluation of him. It will not be the Republican National Committee or some conspiracy.
|Al Tuchfarber is Emeritus Professor of Political Science at the University of Cincinnati. He founded the Ohio Poll and can be reached at firstname.lastname@example.org