Traditional polls have President Donald Trump losing the general election next week. As we learned in 2016, however, traditional polls can’t be trusted. So, let’s look at a non-traditional poll.
The University of Southern California (USC) this week has published a poll – a poll that in 2016 outperformed traditional polls – suggesting that President Trump is going to be reelected.
The “social desirability” question
The poll is the USC Dornsife Daybreak poll, which is different from USC’s standard poll. USC’s standard poll, by the way, has Biden easily winning, 53 to 43 percent. Not so much with the Dornsife Daybreak poll.
The difference between the two is in the question that participants are asked to answer. Traditional polls ask participants who they intend to vote for. The Dornsife Daybreak poll, instead, asks a “social desirability” question.
The particular social desirability question that pollsters asked here was “Do you think your friends and neighbors are voting for Trump?”
A more accurate way to poll?
According to the pollster, “in all five of the elections in which we tested this question, the social circle question predicted election outcomes better than traditional questions about voters’ own intentions. These five elections were the 2016 U.S. Presidential election, the 2017 French Presidential election, the 2017 Dutch Parliamentary election, the 2018 Swedish Parliamentary election, and the 2018 U.S. election for House of Representatives.”
The pollster also noted that “data from the social-circle question in 2016 accurately predicted which candidate won each state, so it predicted Trump’s electoral college victory.”
So, what’s the prediction this time around using this social desirability question? According to the USC, although Democratic President Nominee Joe Biden is ahead in terms of popularity, by four to five points, “it’s looking like an Electoral College loss for Biden.”
What’s going on here?
The question that has to be asked is “why has this polling approach been getting more accurate results?” A hint may lie in a study that was conducted by CoudResearch and released back in August.
That study found that almost 12 percent of Republicans were unwilling to honestly tell pollsters who they were intending to vote for. The corresponding number for Democratic voters was about 5 percent. This means Republicans are more than twice as likely as Democrats to lie during a poll.
The reason for this is obvious: being a Republican these days is difficult. It could cost you your job, your property, and even your safety. Better to lie to the pollsters, and be honest on election day.
We’ll have to see if USC’s approach is right once again.