Was the Trump wave a polling mirage?

DONALD TRUMP’S poll numbers are slipping in Iowa, and a new national poll is the first in a while to show him trailing. But a closer look at the polling suggests that the Trump wave may have been overstated from the beginning. His sizeable lead has been based largely on the influence of Internet polls. Trump’s summer surge looks far less impressive in telephone polls, and polls of likely voters show his lead was always smaller and is now gone entirely.

Looking across all pollsters and modes of pollsters, Donald Trump leads the field by 10 points, according to the Huffington Post, which averages poll results.  Using only online polls, his lead is even bigger. But narrow the field to just telephone polls, and Trump’s lead over Ben Carson drops to 3 points. Drill down further to phone surveys that talked only to likely voters in the Republican primary, and Carson actually overtakes Trump.

The same dynamic is playing out at the state level. In online polls, Trump is tied with Carson in Iowa, and clobbering him by 26 points in New Hampshire. The telephone polling in both states is markedly different, showing Trump now trailing Carson by double digits in Iowa and leading by half as much in New Hampshire.

Online polls are still relatively new, especially for presidential primaries, but their performance in the 2012 cycle suggests they may exaggerate the rise of surge candidates. Herman Cain, of 9-9-9 fame, owed his entire lead to online polls. Looking only at telephone polls, Cain’s lead shrank from 14 points to just 1 point. The same dynamic was present with both Rick Perry and Newt Gingrich, though to a lesser extent.

Fast forward four years, and online surveys make up a much larger share of national primary polling and once again are producing different results. This has had the effect of further boosting Trump, since online polls are now playing a bigger role in the overall average used by poll aggregators. So far this cycle, nine online polling outfits have conducted almost half of all presidential primary polls. In the lead-up to the 2012 contests, there were only two online pollsters producing about a fifth of all national polling.

At the state level, there is even less of a historical record for online polling. In 2012, there were no state online primary polls in Iowa or New Hampshire, one in Florida, and two in South Carolina. The evidence from 2012 is just too thin to say much of anything about the likely accuracy of online, state-level primary polls.

One potential explanation is that online polls may be sampling more non-voters. These non-voters may not be paying close attention and may simply select the candidate closest to the top of their mind, which could be Trump because of his heavy media coverage. This top-of-mind phenomenon might also explain why Trump is doing better in phone polls of registered voters than in polls of likely voters.

It’s too early to know which set of polls will prove most accurate, and impossible to know nationally, since there is no national primary against which to measure. When the dust settled in 2012, the phone polls did pretty well in predicting the final margins in the early states where most of the polling is currently happening. The polling world is in a serious state of flux, but nothing appears to have changed so systematically as to suggest phone polls will be way off this time around. But the state contests will give us a good yardstick with which to measure accuracy.

For now, the evidence, both from this year’s polls and from 2012, suggests the Trump bump was smaller than it appeared.

Steve Koczela is the president of the MassINC Polling Group, a subsidiary of MassINC, which publishes CommonWealth.