If you want to see some key symptoms of unconvincing journalism about social science, look no further than this New York Times piece on the effect of unemployment on families.
Hallmark Number 1: Duh-uh. Job loss has a psychological effect on families. I did not know that!
Hallmark Number 2: Irrelevant statistics. Michael Luo's piece reports on a recent study that "found that children in families where the head of the household had lost a job were 15 percent more likely to repeat a grade." And does this mean that job loss causes children's problems? Or that job loss occurs more often in families with troubles that also cause children to repeat a grade? Without knowing that, it's impossible to say whether this study, or the others cited, supports the article's point.
Hallmark Number 3: OTOH-BOTOH—Halfway into a piece that tells us job loss is a psychological disaster for families, Luo goes into reverse, and we enter the land of "On The One Hand, But On The Other Hand." He writes: "Certainly, some of the more than a dozen families interviewed that were dealing with long-term unemployment said the period had been helpful in certain ways for their families." When a newspaper article makes a claim, but then concedes that, in fact, the opposite is also true, it's a sign that either (a) some editor has put her/his oar in or (b) everyone in the process has realized that the story doesn't justify its main idea.
Hallmark Number 4: Particular people in the story don't match general traits in the argument. That's the curse of the "anecdotal lead" approach, which requires that we writers attach a general observation to a particular person. When it works, it gives abstraction a local habitation and a name, helping people understand the point. When it doesn't work, the idea is bent to match the story, and the story is bent to match the idea. In this case, for instance, the point about the unemployment's psychic pain doesn't fit the lady who says unemployment brought her closer to her children. But more striking is the father in the anecdotal lead. Their family has been stressed, but it didn't experience the most important consequence of job loss, which is not being able to pay bills (hardly irrelevant to the findings about impacts on children in the studies mentioned). The man, Paul Bachmuth, was made to represent a general category of people to which he didn't belong—a fact he himself has pointed out.
Why are there so many stories like this about psychology? Part of it is inherent to the industry. Deadlines have to be met, travel budgets justified. To paraphrase Donald Rumsfeld, you don't write the story with the people you'd like to include, you write it with the people you did. But I suspect that there's also a problem with the way we journalists represent knowledge. Published in magazines and newspapers whose other sections are all about "the facts," our science stories always need to claim that we know something for sure. Yet the real lives we report on are seldom clear examples of some easily-summarized thesis; and the social-science we report on is almost never as certain as we imply. In an information economy, journalists face the same pressures as other workers. All too often, we don't permit ourselves to say "I don't know."