In one experiment, subjects read two documents from 1995 about whether the U.S. should intervene to prevent the sale of fighter jets from Belarus to Peru -- a real debate from this era -- one from the State Department supporting intervention and one from the National Security Council opposing it. Some participants were told that the State document was "classified" at the time the decision was made, others were told that it was the NSC that was secret. "On average, the judgment of information quality when it was secret was significantly greater than the judgment of information quality when it was public," they write.
As I wrote back in 2010 at the height of the WikiLeaks debate, this is a cognitive trap: just because information is classified, does not mean it is valuable.
Photo by Flickr user corporatemonkey.