Neil Irwin in the NYT's The Upshot:
Did unemployment get better or worse during Ronald Reagan’s presidency? In a 1988 survey, some 80 percent of dedicated Republicans accurately said it had improved, compared with 30 percent of loyal Democrats. In the 1990s, the pattern reversed on a range of factual questions about economic and fiscal issues. In a 1997 survey, for example, Republicans were far less likely than Democrats to acknowledge that the budget deficit had declined during the Bill Clinton administration.
As an economics writer, I see the same thing anecdotally. When I wrote articles recently about the unemployment rate’s dip to 5 percent, I received vehement responses from conservatives convinced that the Obama administration was cooking the numbers. They were not so different from responses I received from liberals when the jobless rate was at that level in 2005, during the George W. Bush administration.
In other words, when you ask people about the economy, the answers are less a statement of objectivity and more like what they’d say if you’d asked which pro football team was the best. That has important implications for democracy. How can people judge whether a party is effective if there is no sense of objective truth? And it could even have implications for the economy itself if, for example, conservative-leaning business executives freeze hiring or investment when the president doesn’t share their politics.
But new research from two teams of political scientists adds a wrinkle to these findings. It turns out that the partisan bias in how people answer factual questions about the economy is diminished by this one weird trick: Pay people.
That is a conclusion reached in two new papers in The Quarterly Journal of Political Science, one from four scholars led by John G. Bullock at the University of Texas at Austin, the other by Markus Prior of Princeton and two colleagues.
When survey respondents were offered a small cash reward — a dollar or two — for producing a correct answer about the unemployment rate and other economic conditions, they were more likely to be accurate and less likely to produce an answer that fit their partisan biases.