Many times, as others have correctly noted here, the top-line results of polls in Presidential elections are meaningless. These are the ones that say “Biden leads Trump by 7 points in _____.” First, often one has to look inside the polls as many questions, depending upon the wording, lead respondents to a particular answer. The demographics of the polls are also important as in some Democrats are over-sampled. In some cases, like those from Wisconsin, representative samples of the urban/suburban/rural divide are not accurate. For example, if 45% of a state is considered “rural,” then 45% of respondents should “rural.” Often, we do not see this and it creates an inaccurate assessment expressed in those top-line pronouncements.
Looking inside polls, it is also important to determine the relative strength of conviction for one candidate or another expressed by respondents. That is why they have gradations like “Strongly support,” “somewhat support,” etc. And it goes without saying that pollsters are working under an important, but often incorrect assumption: respondents are telling the truth, if they even answer the call.
Finally, as has been noted, polls were notoriously wrong in 2016. If we went solely by those polls, Hillary Clinton would be running for a second term right now. We are assured by the pollsters that this time they are more accurate, that they got their s*[email protected] together and they know what they are doing. But, pollsters are part of the media which has lost all semblance of confidence and stature they may have once possessed. You cannot really count on aggregator sites like RCP averages, or Nate Silver’s Fivethirtyeight since they are reliant upon what is fed them. The latter gives “grades” to pollsters as if that is supposed to lead us to believe the poll is more accurate from one grade “A” than one graded “C.” Whatever…
Instead, as everyone knows, the outcome of the election comes down to a handful of states that fluctuates from cycle to cycle, but there is a core. This year, who would have thought that Minnesota would be in play for Trump, or that Arizona could be in play for Biden? Wisconsin, Michigan, and Pennsylvania were unnoticed until Trump won them in 2016.
This year, we have 13 states of interest: Arizona, Florida, Georgia, Iowa, Michigan, Minnesota, Nevada, New Hampshire, North Carolina, Ohio, Pennsylvania, Texas (to a lesser extent), and Wisconsin. This writer found it suspect that both Nevada and New Hampshire has a shortage of polls in 2020 (a total of 14 between them) while there are 12 polls out of California. It leads one to believe that pollsters are suppressing some of their results that show Trump putting these states seriously into play.
Most important is the trends in polls- is Trump heading up or down in these 13 states over succeeding months. This could be interpreted as building momentum. In the period from the end of July to the end of August, among 12 of these states (there were no polls out of Nevada in August), Trump averaged a 0.53 bump in swing state polls, although he trailed in most. Still, a half-point increase on a monthly basis is good news despite the depth of the top-line deficit (if any). It shows movement in the right direction.
Through the first 12 days of September, as the general election campaign swings into high gear with Trump on the trail and Biden hiding in his basement, Trump is showing a collective average bump in swing state polls of 0.68 points. In other words, as Trump is more visible on the campaign trail, his poll figures rise. This is the conundrum in which Biden and the Democrats find themselves. They know that if they put Biden out there on the trail, it will likely backfire. Instead, they have opted for scripted, low-to-no turnout events. Biden is relying on digital and MSM advertising, the occasional social media message, and the occasional poking his head out to see if there is a shadow.
What does this indicate at this point? It means that except for possibly Michigan, Trump- if the trend continues- is on track to win the states he won in 2016. In states where things were somewhat dicey- mainly Texas and Arizona- they are moving back towards Trump. It also means that Trump has expanded the electoral map into Minnesota, Nevada, and New Hampshire which are states he lost to Clinton in 2016. In short, Biden has to play a lot more offense than Trump has to play defense and the question is whether Biden is up to that task? The answer now appears he is not which explains why people like Sanders are advising Biden to go full socialist while his handlers hide him in Delaware.
Finally, as has been noticed by others, the RCP average as of Election Day showed an interesting trend. In 2016, Trump’s electoral outcome averaged 2.2 points better than the polling average while Clinton’s final vote total averaged 1.45 points below her final RCP. In other words, an almost 3% deviation between the RCP average and the final vote totals broke in favor of Trump. Considering the power of incumbency and a weak candidate in Biden, what we are seeing in these top-line figures is not reality.