Out of Control Polls

By Jim Ellis

Oct. 3, 2016 — There likely has not been a Senate race polled more extensively than the North Carolina contest between Sen. Richard Burr (R) and former state Rep. Deborah Ross (D), but this week’s survey set literally defies explanation.

According to three different polls, Sen. Burr has opened a wide lead over Ross, or the reverse is evident, or possibly the third scenario, resulting in the two becoming virtually tied, is actually the accurate alternative.

Based upon new Greenberg Quinlan Rosner Research survey numbers (Sept. 10-19; 400 likely North Carolina voters as part of a 1,600 person sample from four states), Sen. Burr now leads the contest by an extraordinary 16 points, 46-30 percent. Keep in mind, this is no planted Republican poll. GQR is a Democratic firm that released an August poll actually projecting Ross as the race leader.

During the same time frame, Republican pollster Gravis Marketing, in a poll we covered earlier in the week (Sept. 23; 624 likely North Carolina voters), surprisingly gave Democrat Ross her largest advantage of the election cycle, 48-39 percent. Also during the same polling period, High Point University Research Center (Sept. 17-22; 404 likely North Carolina voters) sees the senator and Ross in virtually a tied situation, with Burr leading only 45-43 percent.

Sometimes a close race yields strange ballot test spreads, but rarely do we see three such divergent results almost simultaneously. The preponderance of polling suggests that the High Point study is probably the one closest to the actual race status, since so many surveys have found this campaign residing within the polling margin of error.

Pennsylvania, too, is another situation where different pollsters are projecting different leaders. Six different pollsters were in the field during the Sept. 12-25 period, and arrived at varying conclusions. Two pollsters, CNN/ORC and Gravis Marketing project has Democrat Katie McGinty ahead, while the Morning Call/Muhlenberg and Mercyhurst Polling found Sen. Pat Toomey (R) holding a scant one-point edge. Harper Polling forecast the race as a flat tie. This is yet another close race that has been flip-flopping leaders for the past several weeks.

Finally, is a poll that has to be considered the biggest outlier of the election cycle, now becoming public from the Colorado Senate race. The contest between Sen. Michael Bennet (D) and El Paso County Commissioner Darryl Glenn (R) has never been close, and Republicans have long ago conceded this seat. But, a new Gravis Marketing survey (Sept. 22-23; 799 likely Colorado voters) actually finds Glenn taking a small lead over Bennet, 45-43 percent.

The Gravis result is on the heels of another poll, from CNN/ORC (Sept. 20-25; 784 likely Colorado voters) that stakes Bennet to the type of advantage that he has enjoyed for virtually the entire election cycle. The CNN data sees a 53-43 percent spread in the senator’s favor and is closely in line with the all other surveys taken here.

Numbers such as the aforementioned remind us that polling is an inexact science, and that even surveys conducted during the same sampling period can produce drastically different results. In at least North Carolina and Pennsylvania, it is becoming apparent that the only truly accurate survey of these such campaigns will be the final vote totals once they are tabulated on Election Night. This simply means they are currently “too close to call.” Furthermore, it is more than likely that the Colorado poll favorable to Glenn is a clear outlier and an anomaly.

Leave a Reply

Your email address will not be published. Required fields are marked *