I went back to the original paper, and discovered the scientists bending the facts a bit themselves, probably reflecting their own beliefs.

Example from the article (not the paper):

^Kuklinski's welfare study suggested that people will actually update their beliefs if you hit them "between the eyes" with bluntly presented, objective facts that contradict their preconceived ideas. He asked one group of participants what percentage of its budget they believed the federal government spent on welfare, and what percentage they believed the government should spend. Another group was given the same questions, but the second group was immediately told the correct percentage the government spends on welfare (1 percent). They were then asked, with that in mind, what the government should spend. Regardless of how wrong they had been before receiving the information, the second group indeed adjusted their answer to reflect the correct fact.^

One percent? That raised my eyebrows. Out of a $3 trillion budget, 1 percent is only $30 billion. Even if you scrub "entitlements" out, there is no way that means-tested aid for the poor comes in that low. So where did that number come from?