Over the last couple of days, a few of my RedState colleagues have been approaching me in panic-mode to ask about the current polling we are seeing coming in. While the number of polls showing a Biden double-digit lead over Trump may have some freaking out, I am here to tell you, don’t panic. First, National polls after October 1st are liquid hot garbage. They are literally worthless. Second, history is a good indicator that things can change rather quickly, and by quite a lot.
The history to which I refer is 2016.
Beginning as far back as August of 2016, NBC News released the results of a poll that showed Clinton had a 10 point lead over Trump. That same poll was an opt-in online survey done by SurveyMonkey and relied wholly on the responses from participants to determine ideology and propensity, which leads to wild inaccuracies.
McClatchy and Marist had another poll showing Clinton with a 15 point lead, showing party ID at 36% Dem, 37% Independent, and 27% Republican. When you oversample Dems/undersample Republicans, of course, you are going to have a lopsided result. To add insult to injury, the ideology scale used gave a 50-to-37 advantage to Democrats over Republicans.
Quinnipiac had a poll in August as well, showing Clinton with a 10 point lead. The methodology for the poll is no longer available, but Quinnipiac’s last poll of the 2016 race had Clinton ahead by 6 points nationally.
While these polls were taken in August, it was the polls that were taken at the same time in October 2016 that show there’s really no panic here.
If we look to RCP’s 2016 polling page, you can see that numerous polls got it wrong in October 2016. Of note, and highlighted in yellow, are the polls that showed Clinton had a double-digit lead in the last month leading up to Election Day 2016. Ten of those twenty-six polls taken from October 5th to October 21st, showed Clinton with that sizeable lead, while only 2 showed a Trump win. Of the other highlighted green polls, those were showing a Clinton lead outside of their margin of error from what the final electoral result was. In other words, those polls have a measurement of accuracy that is a window in which they think their data will fall. A margin of error is the amount in either direction the poll results could fall and still be “accurate”.
The Reuters Poll for October 20th is a great example. The poll shows Clinton with a six-point lead, with a margin of error of +/- 3.3. That means simply that that poll would still be “accurate” if Clinton won the election within the confidence interval of 2.7% and 9.3%. The larger the sample and more accurate the methodology for the poll, the smaller the margin of error, and thus the confidence interval, will be.
While some will say that Comey’s announcement of the new Clinton investigation on October 28th, 2016 was what changed her polling numbers, in the week before the announcement, numerous polls showed a narrowing of the race, and those numbers didn’t change after the announcement either.
The red line in the above image represents the day that Comey made his announcement. Some polls decisively ended before the announcement and then there were several that were conducted before the announcement but didn’t conclude until after the announcement. The first poll conducted and concluded after the announcement was the Reuters poll on October 29th. While there was a small narrowing after the announcement, it wasn’t what narrowed the double-digit lead Clinton held earlier in October.
Currently, the polls we are seeing from the MSM are showing Biden with a lead in the polls nationally. Of the nine polls taken since October 1, six of them show Biden with a double-digit lead. I’ve already reviewed a number of these polls and will continue to over in VIP, but I can’t imagine that this trend of double-digit leads continues much past the 15th of October.