The Polls Were Wrong Again

By Lionel Parrott – Re-Blogged From Liberty Headlines

During a contentious press conference yesterday, President Trump blamed the media for touting polls that may have suppressed the vote.

The press conference was held following an election which saw the Democrats take control of the U.S. House of Representatives for the first time in eight years. At the same time, Republicans made gains in the U.S. Senate.

“I’ll give you voter suppression,” the president said following a question from a reporter concerned about potential hurdles to participating on Election Day. “Take a look at the CNN polls, how inaccurate they were. That’s voter suppression.”

Presumably, Trump referred to a CNN poll of the generic congressional ballot – where voters are asked whether they plan to vote Democratic or Republican to represent them in Congress.

In comparison to other polls of that measure released in the last week of the election, the CNN poll stands out. The news network’s survey showed Democrats enjoying a massive 13-point lead over Republicans.

But other polls gave Democrats a smaller advantage. The final average of polls on RealClearPolitics showed Democrats with a much smaller, but still significant lead of 7.3 percent – making the CNN poll an outlier. (The actual advantage for Democrats right now is 4 percent, with that lead likely to grow with thousands of ballots left to count in California.)

It’s not just CNN that had a poll that failed to reflect the actual results. Even the president’s favorite network, Fox News, erred in favor of Democrats when it came to an important contest – the U.S. Senate race in Indiana. Fox showed Democrat incumbent Joe Donnelly leading by 7, with the RealClearPolitics polling average being 1.3 percent.

In the end, Donnelly fell to Republican challenger Mike Braun. The margin? 8.4 percent.

The Indiana result shows that even averaging the polls together failed to point to the actual outcome. And it wasn’t just Indiana. A similar situation happened in Missouri, with the polling average giving Republican Attorney General Josh Hawley a 0.6 percent lead over Democratic incumbent Senator Claire McCaskill. Hawley’s margin of victory was 10 times that – 6 percent.

In Nevada, a state where polls consistently underrate Democrats, the RealClearPolitics average of the competitive Senate race there showed a tie, but the Democrat won by 5. And in Texas, Senator Ted Cruz’s reelection over media darling Beto O’Rourke was even closer than what the surveys were predicting.

But in Tennessee, the polls underrated the Republican candidate for Senate, Marsha Blackburn. The polling average showed her up 5; she ended up winning by 11.

Polls were also off in the Florida governor’s race, where Democratic nominee Andrew Gillum was seen as the favorite. But while the polling average had him up almost 4 points, Gillum lost to Republican Ron DeSantis. Current results show DeSantis’s margin as being less than a percentage point.

Individual polls did worse. For example, Quinnipiac University showed Gillum with a massive 7-point lead on DeSantis. The same poll showed Senator Bill Nelson with an identical 7-point lead on Rick Scott.

And a similar dynamic was in play in Georgia, where a poll released by a Democratic organization on the Georgia governor’s race was almost as bad. It showed Democrat Stacey Abrams with a 4-point lead on Republican Brian Kemp.

Not surprisingly, races where few polls were conducted were the most likely to prove surprising on Election Night. Most observers, for example, considered the campaign to unseat Democrat Joe Manchin in West Virginia to be a lost cause – perhaps because of a poll from “Strategic Research Associates” in mid-October showing him with a gigantic 16-point lead.

Possibly, the race tightened since then – Manchin won by only 3 percent.

And, in the aforementioned Missouri Senate race, two organizations released polls just before the election, both of which showed Democrat Claire McCaskill with a 3-point lead. Given that she lost by 6, that amounts to missing the margin by 9 percent, quite a significant miss.

One of the same polling firms that put McCaskill in the lead also put out a poll showing Democrat Jon Tester of Montana with a very comfortable 8-point lead. While Tester survived, he won by only 3 percent. If the poll was accurate, the only explanation is that the Republican candidate won virtually all of the undecided voters.

Overall, polls for the U.S. Senate and for governor’s races generally underestimated Republican candidates. It is unknown how much of that was due to sampling error, voters breaking late, or attempts at suppression.

But the lowered expectations prior to Election Night, compared to the actual results, may have contributed to a sense among Republicans that the night went better than expected – making the “blue wave” feel more like a “blue trickle.”


4 thoughts on “The Polls Were Wrong Again

    • It (should be) al statistics. Given a population size, if you select samples at random, then X number of samples gives you confidence within a range. Therein lies the problem. In the US, if you select your sample voters mainly from cities, then you’ll be over-represented by Democrats.

      The output could be off due to poor poll design, or it could be due to intentional bias. When the polling groups favoring the Ds all overestimate the Ds chances, it’s hard to call it anything but bias.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s