Does A Classification Marking Inherently Bias Perceived Reliability?

In “HOW SECRECY CAN DISTORT DATA” (, David Berreby cites two studies that posit that an individual will rate classified information, on average, with 15% more credibility than non-classified information.  I find the article and the studies cited to be naive in their approach to supporting the notion that adding a classification label lends some inherent credibility to information when judged by legitimate professionals.

The methodologies of these studies don’t exactly lend themselves to authoritative results. Perhaps if individuals from the Intelligence services were recruited for comparison it may be slightly more informative but even within those groups the ability to discern credibility (and the responsibility to make that judgement) run a very broad spectrum. Further, gauging between classified and unclassified sources is probably not meaningful as decisions are made from multiple lines of evidence in _any_ field meaning that a bias, should it actually exist, would likely not be a factor in real-world decisions. I would be very interested to see a study performed with a more valid population and a measure inserted to test if these biases actually influenced any decision in a meaningful way.

Anyone know of any studies analyzing these factors?