Brian C. Joondeph:
“…Presidential pollsters have short memories. It was a mere four years ago that most pollsters faceplanted after predicting a Hillary Clinton landslide victory. Rather than humility and introspection after such wildly inaccurate predictions, they have doubled down and may be falling into the same trap of wishful thinking as they did in 2016.
Opinion polls are only as good as their samples. Assessing President Trump’s popularity in Boulder versus Sturgis would yield vastly different results based on the population being surveyed.
A recent NBC News/Wall Street Journal poll conducted Sept. 30 to Oct. 1 showed Joe Biden with a 14-point lead over President Trump, 53 to 39 percent, up from an 8 point lead a week earlier. That’s it then, game over. Time for Biden to start packing for his move from his basement to the White House.
Who was polled to give Biden such a commanding lead? Looking at the poll internals, something the media prefers to hide, reveals that registered, not likely, voters were surveyed, one strike against the poll. As elections move closer to election date, that calls the sample into question. Given that voter turnout for presidential elections is typically just over 50 percent, half of those surveyed won’t vote, regardless of their opinions.
Of the 800 registered voters surveyed, Democrats were oversampled by 9 points, 45 to 36 percent, explaining much of Joe’s supposed 14 point advantage.
This poll echoes a July Quinnipiac poll finding Biden opening a 15-point lead over Trump. Again this was registered voters and oversampled Democrats by 10 points, 34 to 24 percent, creating much of the 15-point lead.
Are there 8 to 10 percentage points more Democrats than Republicans out there? If so, then why isn’t Hillary Clinton the one running for re-election? In reality it may be the opposite. Politico reported Republican voter registration surging in Florida, 41 percent more new Republican voters compared with Democrats…”
