US election

Polling Comment: What to Believe?

By October 26, 2016 No Comments

by Ron Faucheux, President of Clarus Research Group, MSLGROUP’s full service polling and research company

Many have asked about the tracking polls produced by Rasmussen and IBD/TIPP, which are showing Trump doing better relative to Clinton than most other polling. Here are a few observations:

Rasmussen’s surveys are conducted through a combination of automated phone calling to reach voters with landlines and an online survey tool to reach voters without landlines. Generally, these are the least expensive interviewing methods available.

The final poll Rasmussen reported before the 2012 election had Romney winning the race by 1 point. Of course, Romney lost––and Obama won by a 3.9-point margin. That wasn’t the only time in 2012 when Rasmussen’s results were out of line with other polling. In mid-September, for example, Rasmussen showed Romney ahead by 2 points, while six other national polls taken close to the same time showed Obama with an average 4.2-point lead.

The IBD/TIPP poll is based on live phone interviews conducted via cellphones and landlines. It calls itself “The Most Accurate Presidential Poll in America”––but, apparently, that wasn’t the case in the 2012 election, when their final pre-election poll showed Obama ahead by 1 point, which was nearly 3 points off the actual result (Obama +3.9). The two final reported polls that came closest to the actual results in 2012 were theABC/Washington Post poll, which showed Obama ahead by 3 points, and the Pew poll, which also had Obama in the lead by 3 points. It should also be noted out that the finalCNN/ORC and GWU/Battleground polls in 2012 had the race a tie, the final NBC/WSJ poll had Obama ahead by a point and the final Gallup survey had Romney ahead by a point.

In the 2008 election, the final reported IBD/TIPP poll came close, showing Obama ahead by 8 points, as did the final NBC/WSJ poll. Obama won by 7.3 points. The closest reported poll, though, was the Fox News poll, which had Obama ahead by 7 points. The finalRasmussen poll put Obama on top by 6 points that year. In 2004, the final reportedIBD/TIPP poll put Bush ahead by 2.1 points, which was close to the CBS poll which had Bush ahead by 2 points. On Election Day, Bush won by 2.4 points.

Three things to keep in mind:

  1. Looking only at the final polls before Election Day has become the trendy way to judge the reliability of public polls. But, just because a polling firm gets the last poll right doesn’t mean it was right all along. Even a stopped clock is right twice a day.
  2. Remember that polls don’t predict. As “pictures in time” they only tell us where things were when the interviews were conducted. It is a mistake to treat any survey taken weeks before Election Day, whether it’s in line with other polling or not, as a predictive tool.
  3. There have been other times in recent history when polls bounced around at this point in presidential campaigns. At about this time before the 2012 election, for example, three polls showed Romney ahead, two had Obama on top and one had it a tie. All of the polls, though, showed a close race. Obama ultimately won by 3.9 points. Around the same period in the 2008 election, a Pew poll had Obama in the lead by 15 points, while the IBD/TIPP and the GWU/Battleground polls, taken about the same time, had Obama ahead by only 3 points. Obama ultimately won by 7.6 points.

 

The polling results cited in this commentary are based on the Real Clear Politics archive.