Is Donald Trump’s political fate sealed?
That’s the dominant question as many stories appear about how poorly he is doing in polling which Trump tries to discredit by calling them fake, politically biased polls, even if they’re Fox News polls. A lot of people predict he can’t recover from such bad polling numbers in light of the raging coronavirus, battered economy and continuing protests and his “secret police” reaction to it.
However, polling maven Nate Cohen disagrees. Axios:
FiveThirtyEight founder Nate Silver said on ABC’s “This Week” that while President Trump’s reelection bid is “clearly in trouble” due to his dismal coronavirus approval ratings and polling in swing states, he does not believe the president’s “fate is sealed.
Silver points to some events that could possibly shift the polling and adds:
“We found historically that when there are lots of major news events and economic disruptions, an election becomes harder to predict,” Silver said.
…”So while he’s clearly in trouble, I do not buy that Trump’s fate is sealed. A lot could change in the next 100 days — things could get worse, still, for the president.
“But a turnaround in the COVID situation by the fall could make the election more competitive.”
Historian Aaron Astor offered this analysis on his Twitter account:
Every four years it becomes clear that much of the public has no idea how to read political polls. It's especially bad when so-called experts reinforce bad habits re: poll-reading. Here's hoping that some of those experts learned from 2016. Here are a few things to emphasize:
— Aaron Astor (@AstorAaron) July 26, 2020
1) Not every poll is equal in quality or inherent bias. 538's grading system is helpful but not failsafe. Don't judge a poll to be great simply because it got 1 tough race right without considering how many it got wrong – and vice versa. Good pollsters occasionally get one wrong.
— Aaron Astor (@AstorAaron) July 26, 2020
2) Polling averages are almost always more useful than individual polls. Use RCP or 538 or any other high-quality aggregator and look for trends.
— Aaron Astor (@AstorAaron) July 26, 2020
3) Always check to see if a particular pollster has polled a state before. If so, watch for the trendline in that poll. If not, don't treat a difference between two different pollsters as "tightening" or "widening" unless matched with other polls showing the same.
— Aaron Astor (@AstorAaron) July 26, 2020
4) Be very careful with Party ID. Sometimes a skewed party ID sample suggests non-response bias – a bad news cycle for candidate X means supporters of that candidate are less willing to talk to pollsters at that time. Other demos are more useful (age, education, race/ethnicity).
— Aaron Astor (@AstorAaron) July 26, 2020
5) Don't just look at the margin between candidates. Look at how close the leader is to 50. A big problem in 2016 was misreading polls with lots of undecideds. Relatedly: post-election studies showed these folks really were undecided until the end and not shy Trump voters.
— Aaron Astor (@AstorAaron) July 26, 2020
6) Be very careful with demographic sub-samples. When you see an odd result in a sub-sample, it's probably because the margin of error is ginormous.
— Aaron Astor (@AstorAaron) July 26, 2020
7) 2020 is not 2016. It's not 2012 or 2008 either. The dynamics are different each time. The makeup of the undecideds is different each time. The importance of third parties is different each time.
— Aaron Astor (@AstorAaron) July 26, 2020
8) One of the striking things about 2020 is how stable the race has been so far, esp. compared to 2016. All the more reason to treat an individual poll outside the trend (in either direction) with caution. Not to ignore it outright, but treat with caution.
— Aaron Astor (@AstorAaron) July 26, 2020
Conclusion: We are 100 days from the Election. In reality, we are much closer than that with early voting. Big events to come are Biden's VP choice and debates. No "virtual" convention bounces likely. The pandemic is still THE biggest issue. It has health AND economic effects.
— Aaron Astor (@AstorAaron) July 26, 2020
Joe Gandelman is a former fulltime journalist who freelanced in India, Spain, Bangladesh and Cypress writing for publications such as the Christian Science Monitor and Newsweek. He also did radio reports from Madrid for NPR’s All Things Considered. He has worked on two U.S. newspapers and quit the news biz in 1990 to go into entertainment. He also has written for The Week and several online publications, did a column for Cagle Cartoons Syndicate and has appeared on CNN.