The poll is largely in line with the average poll since April that puts Biden 6 points ahead of Trump nationwide.
The stability in the polls is record-breaking. Biden’s lead is the most stable in a race with a running race since at least 1944. This could mean that it will be more difficult to change the trajectory of the race going forward, even if this remains more than close enough to easily win both candidates. .
To find out, I went back and observed how all national polls diverged between January and early May of the election year.
This year, 95% of all individual polls so far have shown a result within 6 points of the average. This is basically what you would expect if you took a lot of polls and the race didn’t move (i.e., the only changes are statistical noise from sampling). It’s a ridiculously historically speaking range.
The previous low for a similarly constructed 95% confidence interval was 8 points (2012). The median cycle had a 95% interval of 13 points from the survey average. In other words, about double the range of polls we’ve seen so far in 2020.
Sometimes, of course, the range of results can be even wider than the median cycle. Lyndon Johnson had a roughly 35 point lead over an over 60 point lead over Barry Goldwater in early 1964. Jimmy Carter opened 1980 with a 30 point or better lead in some of the Ronald Reagan polls. By early May, his lead had gone down to a single figure and he even followed Reagan in some polls.
In fact, there appears to be a relationship between the initial survey movement and how well the initial survey predicts the results. In cycles with smaller than normal early movements, the average difference between the surveys during those four months and the final results was 5 points. In cycles with larger than normal early movements, the average difference between the surveys during that period and the final result was 14 points.
In other words, it would adapt to previous models if surveys are generally close to results right now. This corresponds to what you would expect, given that more voters have a strong vision of Trump than any other incumbent since at least the 1980s (when the first question was asked), as I have already noted.
The obvious catch here is that a 5-point difference between where the polls were and the November result could be huge this year. Biden rose by only 6 on average nationwide. A 5-point shift in Trump’s direction would mean a 1-point Biden victory in the popular vote. This could leave him on the wrong side of a popular vote and the split of the constituency. Remember, Trump lost 2 points nationwide in 2016 and still got a win in the constituency.
And we’re just talking about the media here. The 1956 campaign, which initially saw limited rounds, caused Democrat Adlai Stevenson to do 10 points more than his first poll.
The bottom line is that while stability in the 2020 run so far projects minimal movement from here on, it in no way guarantees Biden a win. There is still a range of possible outcomes.
Coffee enthusiast. Travel scholar. Infuriatingly humble zombie fanatic. Thinker. Professional twitter evangelist.
