Just before the election, Robert Cahaly of Trafalgar Group polled several key swing-states and released results showing Donald Trump winning.
He was ridiculed.
“When was the last time you saw someone run against the supporters too?”
His Pennsylvania and Florida polls showed Trump poised to win tight contests. And, well, the Democrats and the NeverTrumpers were certain he was wrong.
“These crosstabs by Trafalgar Group are hilarious,” wrote conservative writer Marcus Hawkins on Nov. 7, the day before the election. “First Florida, now almost 1/3 of (Pennsylvania) Blacks back Trump. Yeah, no.”
Cahaly predicted Trump would win Pennsylvania, with Trump at 48.4 percent and Democrat Hillary Clinton at 46.5 percent. The Republican nominee ultimately won that contest 48.8 percent to 47.6 percent, leaving Cahaly off by just fractions of a percent.
He also lined up Ohio, Florida and Michigan into the Trump column. Trump won them all.
Trump would also win the Electoral College, Cahaly said. It happened.
"Our projection: Trump 306 Clinton 223 w/CO, VA & WI possibly breaking Trump as 'hidden voter,'" Cahaly tweeted.
Again, he was ridiculed, especially on Twitter. But his prediction was almost perfect. Trump is on the path to win 306 electoral votes once Michigan is called. With 100 percent reporting Trump leads Clinton 47.6 percent to 47.3 percent in the Great Lakes State, according to the New York Times.
His caution on Wisconsin also proved prescient. Wisconsin indeed had shy or hidden voters. They gave the state to Trump.
Since Election Day, the media has been asking why their polls and other experts -- especially statistician Nate Silver of FiveThirtyEight -- got it so wrong. Silver, by the way, rated Cahaly a C-grade pollster.
The New York Times admitted their reporters had been chattering to each other in Manhattan, and didn't hit the road enough to talk to farmers and nurses.
And on Tuesday morning, Silver's site gave Clinton a 72 percent chance of winning.
After the dust settled, some have also been calling Cahaly, a Georgia political consultant, asking what he did right. Much of his method was covered by LifeZette on Nov. 2 .
Cahaly told LifeZette Friday his most unorthodox poll question -- the neighbor test -- proved to be an accurate forecast method.
Cahaly says he did extensive research to find out why Trump was so resilient in polls. He added in voters to his polling "universe" that has been inactive for years. Some had not voted since 2006. Cahaly said many pollsters ignored them.
But the main "secret sauce" Cahaly used was to ask the "neighbors" question.
The neighbor question was a test of a theory -- the "shy Tory" theory.
That theory is based on polling in the United Kingdom in 1991 and 2016. Then, pollsters found conservative voters were shy about telling pollsters how they would vote. It happened again this year when the United Kingdom had a referendum on exiting the European Union, also known as the "Brexit" issue.
There was heavy media pressure on Brexit supporters. Portrayals of Brexit supporters were demeaning. Brexit supporters were cast as isolationists, racists and "swivel-eyed loons."
So of course some didn't tell pollsters they were for it. On the eve of the Brexit vote, Bloomberg pollsters found a plurality of Britons -- 46.2 percent -- supported remaining in the European Union. Only 44.3 percent supported Brexit.
Brexit won the referendum.
Cahaly said the same thing happened this year. Democratic presidential nominee Hillary Clinton not only ran against Trump, she attacked his supporters. She and surrogates such as Bill Clinton called Trump supporters deplorable, rednecks and racist.
So U.S. respondents were shy about saying they would vote for a candidate or cause that has been tagged as racist or distasteful by the media, by the Democrats or by popular culture and loud celebrities.
"When was the last time you saw someone run against the supporters too?" Cahaly asks.
To pry these shy voters out into the open when contacted by phone, Cahaly did two things.
One, he used a robotic dialer and software to poll. People felt freer to talk to an impersonal, non-judgmental computer.
Second, his questionnaire asked who the respondents' neighbors would vote for. It was a psychology that Nate Silver didn't use or count on. If people wouldn't admit they would vote for Trump, they would admit their neighbor would.
This would give Cahaly a secondary idea of Trump support. Cahaly figured the Trump surge was in between his actual "Trump versus Clinton" results and what the neighbor number was. It was usually much higher for Trump.
In using psychology combined with normal polling science, he proved the shy Trump voter existed and he became one of the most accurate pollsters on Election Day.
Before the election, Cahaly told LifeZette he welcomed the results to prove his theory, and that he expected a hidden tsunami of Trump voters, more than a few of them long inactive.
By 8 p.m. Tuesday, results in "blue states" that hadn't voted for a Republican presidential candidate since the 1980s, started to pour in.
News panels from CNN to Fox News to NBC news looked stunned. They did not know what was happening.
The Times and others should have looked at Cahaly's website -- on their way out the door to try to talk to voters far from Manhattan and the Beltway.