Pollsters making same mistake as in 2016?

Brian C. Joondeph:

While the moniker “fake news” is typically reserved for cable news and some of the more prominent newspapers in America, the term could also be applied to presidential polls. How many pollsters predicted Hillary Clinton would win the 2016 election by a landslide up to and including the day of the actual election?

The “paper of record,” the New York Times, told readers on election day, Nov. 8, 2016, that Hillary Clinton had an 85 percent chance of winning the election. Their prediction was updated on 10:20 PM ET, oblivious to the evening smiles on cable news shows that were curdling into frowns and outright hysteria.

Screen shot from the New York Times web site on Nov. 8, 2016 // Fair Use

Not only were they wrong, but they were also in abject denial over events unfolding before them in real time. The NY Times was in good company as most pollsters got it wrong in 2016, with Rasmussen arguably the most accurate of the bunch.

Have pollsters learned from their mistakes? Or are they about to step in it again in 2020? Judging by some recent polling news, it appears that history may be repeating itself. As the old proverb says, “Those who do not learn from history are doomed to repeat it.”

Several days ago, this headline ran, much to the delight of CNN and MSNBC anchors, “Fox News poll shows Trump losing to Biden, Warren, Sanders and Harris.” That’s it, then, the race is over. The message to Trump supporters is to give up in despair and leave those MAGA hats in the closet.

The fact that this is a Fox News poll should throw cold water on the narrative of Fox being in the tank for President Trump. Fox News didn’t conduct the poll, however, but instead outsourced it to Braun Research Inc as noted in the poll’s methodology.

Opinion polls are only as good as their samples. The Fox News poll surveyed registered voters, unlike Rasmussen, which surveys likely voters. Why is that important? CNN reported that only 55 percent of eligible voters actually voted in 2016. An opinion poll that queries the half of the population that doesn’t even bother to vote will be far less accurate than one surveying those most likely to vote.

...

There is something else that is likely skewing the polls.  The Democrats and the media have so demonized Trump and his supporters it is unlikely that those voters will tell pollsters who they are voting for.  The reason Rasmussen was closer to the mark last time was not only that he was polling likely votes, but he was doing it with a polling method that gave the voter absolute anonymity.  Those being polled did not have to have a conversation with the pollster.  In a climate where wearing a MAGA cap is enough to trigger an assault by a deranged Democrat, those things are important.