Washington (AFP) - Political polling around the world is often accurate, despite the shock victory of US President Donald Trump who was widely predicted to lose the 2016 election to Hillary Clinton, researchers said Thursday.
Polls correctly predict elections nine out of 10 times, said the findings by the University of Houston published in the journal Science.
"This study suggests polling data can be utilized not just in the United States but globally to predict election outcomes," said lead author Ryan Kennedy, a political scientist at the University of Houston's Center for International and Comparative Studies.
"It would be a mistake to abandon the enterprise. The future really is in trying to make better quantitative predictions."
The study focused on elections in which people vote directly for a leader, rather than having that leader elected by parliament.
For one part of the study, researchers used polling data in the weeks prior to elections in Latin America in 2013 and 2014, "and correctly forecast the winners in 10 out of 11 elections, or 90.9 percent of the time," said the study.
A second part involved poll predictions for more than 500 elections in 86 countries in 2013, and gave a success rate of 80.5 percent.
"People normally wouldn't find this surprising," said Kennedy.
"Where there was a poll, it was reasonably good at predicting outcomes, even in places you wouldn't think you'd be able to have accurate polling."
In the case of the US presidential election, former secretary of state Clinton won the popular vote by nearly three million over Trump.
But Trump won the electoral college, with victories in key states that were enough to hand him the top US office.
The failure of even the most sophisticated polling models to predict Trump's electoral college victory "has led to a new wave of criticism of quantitative methods for predicting elections," said the study.
But researchers said it bears remembering that polls are not bulletproof.
"A forecast placing the probability of an outcome at 80 percent will be wrong 20 percent of the time," said the study.
Kennedy recalled that he and his collaborators predicted an 84 percent chance of a Clinton victory.
"That meant a 16 percent chance of a Trump victory," he said.
"Unlikely but still possible."