Polling & Politics

US election answers from pollsters? It’s the undecideds!

Political polling is still an art, not a science. It was evident during an intriguing Webinar August 20th, 2020 from ESOMAR featuring some of the best pollsters currently examining the US Presidential and State by State Elections in the weird and wacky world of Trump and his ‘regime’ where we can always expect the unexpected. 

The potential clues to watch for in determining the results in November – sometime! – appeared clear along with some fascinating dimensions for either ‘likely voters’ or ‘registered Democratic or Republican voters’. 

The various pluses and minuses of the technical research executions can significantly affect any polling results, however kudos to the contributors: Jean-Marc Leger, President Leger; Frank Graves, President EKOS Research Associates; Cliff Young, President, Ipsos Public Affairs, USA; and Chintan Turakhia, EVP, SRSS with Jennifer Agiesta, Director of Polling CNN. 

Turakhia and Agiesta emphasized the importance of fully (in practice over) representing the Battleground States in any National Presidential poll due to their influence on the final results which are based on the Electoral College decision structure State by State in the US and not the total popular vote. (The US is not a Republic in the strictest sense of the word!) SRSS continually re-assess and revise which States are Battlegrounds. Their national monthly 1,000 in-tab sample uses an RDD polling technique, incorporating both land and mobile lines, based on “likely voters”. “Likely voters” for SRSS/CNN are based on intention, interest, and self-reported voting history. It is weighted by broad demographics. 

This research foundation has, in my opinion, paid dividends for the value of the SSRS National results and how much CNN viewers can lean on them versus some other polls reported in the media. Jennifer stated that CNN updated its polling quality standards in 2019. To no surprise, they found that Dems are significantly more likely to vote by mail in this election versus Republicans (versus earlier elections!). It is important to remember that in the USA 5 States, Colorado, Hawaii, Oregon, Washington, and Utah vote exclusively by mail and many others fully allow, or will allow, extensive use of mail/absentee voting. These simple facts surely underline the rationale for the regime’s efforts to kibosh the Postal Service, which will also help to reduce voter turnout. 

Leger stressed the level of turnout as a key results driver noting that Battleground States do, however, have much higher turnouts than others. His findings on the effect of turnout over the years-and by ethnicity – blacks are much higher than Hispanics, and age – older much much higher than younger- echoed that of SRSS. Leger uses “registered and absolutely certain to vote” as his primary sampling base. He noted that, the US has a history of low turnouts compared to many other democracies ranking the US 26th worldwide and that traditional polling techniques likely overestimate turnout. Overall turnout was 53.4% for the Mid-Term elections in 2018, the highest for ten years. In a bold prediction, he suggested that a high turnout in the region of 63% – 64% would give Joe Biden the Presidency. 

So, we have another clue to “The Answer”. The turnout dynamic provides a clear basis for the, “suppress the vote” tactics of the Republicans versus the “get out the vote” of the Democrats. And we all thought it was all about policies, experience, ethics, and demeanour! 

Although very tricky to determine the actual percentage of the population, Leger, I believe correctly, posited that based on the brutal polarization of the current electorate, “it’s the people that stay at home (and did not vote by mail!) plus the undecideds that will ultimately decide the final result”. So, we have another dimension of “The Answer” to watch closely. I do recall that in 2016 the largest number of people in that US election was the people that did not vote, reflecting its relatively low turnout compared to other election years. Probably concentrated among the undecideds?    

As confirmed by Graves, EKOS in researching underlying implications and motivations of voters, “There is a paradox. A horrible outlook on a host of issues but a seemingly close race.” His interpretation of his research findings was that the significant increase in polarization was primarily driven by “authoritarian populism”. He defined this as a “rooted fear and emotionally powerful hostility to outgroups”. I would suggest, beyond sad – but how true. From a US polling perspective with all its complexities, he suggested that national polls should be built from the ground up an area by area, State by State similar to SRSS/CNN’s approach.

Young presented some important potential predictive results that differ substantially from the vast majority of the polling numbers. Ipsos had Trump with a ~40% likelihood for victory versus ‘FiveThirtyEight’ at ~ 28% and some others as low as 5%. This prediction is due to mixed signals from their fieldwork. Ipsos believes that various prior results and trends, notably for sitting Presidents, are an important factor. However, I am not sure I would agree with the emphasis of this element of their predictive model, especially in a time when the unpredictable-unpredictable appears to be the norm, and the polarization is so extensive and entrenched.  

Not surprisingly Young offered research that paralleled the other companies regarding the complete polarization of the US electorate on every main issue, but with the warning. “Don’t count Trump out!” Echoes of 2016? However, this time the poll results have been “very stable for the past ten weeks” unlike 2016. 

This rich, research-based US polling webinar provided some very real clues to, “US Elections Answers”. 

  • Watch the turnout in key voting segments in the Battleground States and especially the undecideds. 
  • The first debate on September 29th could provide an inflextion point.
  • The, “James Comey” unpredictable element can change everything.
  • The final result will be dependent on a relatively very very small percentage of the population.
  • Expect the unexpected!
  • Do not expect results by November 4th!  Mid-November maybe? 

This article is an extended version of a Commentary that appeared in Media Post August 24, 2020

Leave a Comment

* By using this form you agree with the storage and handling of your data by this website.
Please note that your e-mail address will not be publicly displayed.

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Related Articles