What if the polls are right?

24 September 2020

1:35 AM

24 September 2020

1:35 AM

I was wrong, and I apologize. For far too long, pundits have pronounced confidently on matters of national import, pocketed the fee and moved on to the next mercenary opportunity for reckless prognostication and bare-faced self-promotion without so much as a backward glance to see if their opinions were based in fact and their predictions confirmed by events.

On August 26, I foolishly suggested in these pages that by early September, polls would show ‘Biden’s lead over Trump shrinking into the margin of error, and Trump edging ahead in a couple of swing states where he is now behind’. This, I now realize, was wrong.

By September 8, Trump and Biden were still tied on 48 percent each in Florida, according an NBC/Marist poll taken in the first week of September. It took another week for Trump to pull ahead, and a further week for the news to break: in a Washington Post/ABC poll taken between September 15 and 20 and released on September 23, Trump leads Biden in Florida by 51 percent to 47 percent. The same poll finds Trump edging ahead of Biden, 49 percent to 48 percent, in Arizona.

Any fair-minded observer will agree that it was wildly irresponsible of me to predict Trump edging ahead in ‘early September’. Especially in an election season when the news is riddled with fakery, syphilitic with partisanship and about as reliable as Joe Biden’s teleprompter skills. I should have written ‘mid-September’. I was utterly wrong. By about five days.

Nor did Biden’s lead over Trump shrink into the margin of error in early September. Unless you believe Rasmussen’s polls. In early September, when all the other polls had Biden ahead by an average of 6 to 8 percent, Rasmussen had Biden’s lead down to 2 percent.

The Rasmussen national poll that placed Trump a point ahead of Biden, 47 percent to 46 percent among likely voters, was taken between September 9 and 15. Definitely not early September. Again, mea culpa. If I’d got the wrong side of $105,261 and bought an MA in Journalism from Columbia before I started reading the tea leaves, I’d tear up my diploma and demand a refund. Or a teaching job.

We are not supposed to believe Rasmussen’s polls, even though Rasmussen’s polls turned out to be consistently more accurate in 2016 than any others. And even though Rasmussen’s early September poll put Biden ahead in Wisconsin, Michigan and Ohio by more than the margin of error.

So I don’t believe my disbelief about the Rasmussen polls. But I have trouble believing anyone’s polls. The pollsters of the ancient world called this the paradox of Epimenides. What to believe when a Cretan tells you that ‘All Cretans are liars’?

The data flow is corrupted at its source. We are drinking from a river that has a dead sheep upstream. A whole flock of them, in fact, placed there by the parties and biased outlets that commission the polls, the pollsters who word the questions, select the polled and spin the results, and the media who hype them as fact.

As a mathematical illiterate, I draw two conclusions from the latest polls. One is the trend line, the other the trend for lying. The trend line shows that Biden’s national lead over Trump has narrowed since the summer.

The trend for lying takes two forms. One is the pretense that Biden’s national lead equates to victory on November 3. It’s the swing states that count, and the trend line there is very clear: Trump is closing the gap to within the margin of error and even pulling ahead.

The other trend for lying is that no sane resident of a swing state who was thinking of voting for Trump would admit it to a stranger. The social stigma is too great, the professional risk too high. The polled, like the pollsters and the pols, are lying out of self-preservation. What if you tell the neighbors and your colleagues, but your swing state swings blue in November?

Got something to add? Join the discussion and comment below.

Show comments