A Trafalgar Group poll has President Trump taking a slight lead in Pennsylvania, showing the small Georgia research firm is once again going against the corporate-polling grain, which has Democratic nominee Joseph R. Biden with a healthy lead.
The survey of 1,076 likely Pennsylvania voters taken Sunday had Mr. Trump with the support of 48.4%, to Mr. Biden’s 47.6% once “leaning” voters were included. The Trump edge of 0.8 percentage points is, however, less than the survey’s 2.9-percentage-point margin of error.
Trafalgar Group also has Mr. Trump ahead in Michigan and other swing states.
Chief strategist Robert Cahaly says his small team of analysts know how to find Trump-shy voters, a skill displayed in 2016 when it — and Mr. Trump — shocked the media establishment.
In 2016, Trafalgar’s final polls in Pennsylvania and Michigan said Mr. Trump would win these crucial states, even though 2016 Democratic presidential nominee Hillary Clinton was taking them for granted. Every other pollster — including big liberal daily newspapers, college number crunchers and TV networks — confidently said he would lose those two states.
“We knew Monday night that Trump was going to win,” Mr. Cahaly told The Washington Times afterward.
As in 2016, Trafalgar is offering Trump backers a ray of hope against a sea of “blue” poll numbers on the much-watched Real Clear Politics (RCP) average “poll of polls.”
Trafalgar has Mr. Trump ahead in six of seven key swing states. According to the pollster, in addition to his slight lead in Pennsylvania, Mr. Trump is:
⦁ Up by 4 percentage points in Ohio. Other polls have it closer.
⦁ Up by 2 points in Florida, while the RCP average has Mr. Biden ahead.
⦁ Up by 2 in North Carolina against an RCP average of 4.5% for Mr. Biden.
⦁ Up 4 in Arizona against an RCP average of Mr. Biden up 2.
⦁ Up 1 in Michigan while the RCP average is a whopping plus-9 for Mr. Biden.
⦁ Down 1.3 points in Wisconsin, a margin that is still less than other pollsters’ numbers a week before Nov. 3.
Trafalgar discovered through comparison polling that some distrustful Trump supporters would not disclose their choice, perhaps relating to how they are negatively portrayed.
Trafalgar also created a large sample size of 1,000 to find Trump voters who had not gone to the polls in a while. And it also added a “neighbor” question to let the respondent know they were not alone.
The firm relies on six different methods: live callers; automated voice; texts; emails; and two proprietary cybermethods.
Mr. Cahaly says the questioning lasts just 1 or 2 minutes to assuage impatient people who don’t like long surveys.
In 2016, Trafalgar also correctly predicted Trump wins in North Carolina, Georgia and Florida. And pegged his Electoral College performance spot-on, at 306 votes.