Yet another example of an overblown headline based on a relatively anodyne academic study:

Americans have significantly less confidence in their religious leaders than they did a generation ago and more than two-thirds would prefer they not dabble in politics, according to a new book by a Duke University professor.

Chaves' research found that between 1973 and 2008, the percentage of people with great confidence in religious leaders declined from 35 percent to less than 25 percent....

There is a declining, though still very high, belief in God. In the 1950s, 99 percent of Americans said they believed in God; in 2008, about 92 percent did.

First, this drop isn't that large, and could be as little as half the size reported given the confidence intervals. Second, it really isn't surprising. What if they had phrased it this way: "After thirty-five years of constant reports about clerical child abuse, mega-church pastors with mistresses, and reverends with gambling addictions, slightly fewer Americans have high confidence in men that they half-listen to for a few hours a month." Essentially what the article says is, "Americans have never had a particularly lofty view of religious leaders, but lo and behold, there are some fluctuations in that level over time." 

I'm not trying to be too hard on Chaves here--Lord knows academics have heavy incentives to make their findings seem more shocking than they really are. Rather, it is the media's insistence that research must be presented in this way in order to garner attention from the public. This leads to a couple of things that I view as negative. First, the public over time begins to have unrealistic expectations of science, that it should always be novel or exciting and clearly tied to a "useful purpose." Great discoveries have often been so only in retrospect, whereas most day-to-day science is more like puzzle solving.

The second reason that media portrayals like this hurt science is that researchers are tempted to pursue projects that will lead to such supposedly shocking headlines, rather than work out important puzzles that remain in influential theories. This in turn can lead to temptations to falsify or "adjust data" to make sure that the surprising claim is supported, which are then discovered, which then lowers the publics view of science, which makes us want to surprise them with interesting findings, and so on...

Not to despair too much, though. Chaves actually does have a couple of interesting findings:

-- Fewer Americans today approve of their religious leaders getting involved in politics. In 1991, about 30 percent of Americans strongly agreed that religious leaders should avoid political involvement; by 2008, 44 percent felt that way.

-- Religion and politics are more closely intertwined than a generation ago.

"Several decades ago there was not a strong correlation between how religiously active you were and whether you voted Republican or Democrat," Chaves says. "Now, there is. If you're religiously active, you're now more likely to vote Republican. That's a very important development and is part of what leads people to talk about increasing polarization in American society."

See? Not overblown, just simple statements that correspond to facts. I'd be interested in seeing more details on the second point, but not if it's couched in language that's meant to get my blood pressure up.