The statistic that wouldn’t die

oneinfour 01So I’m reading Washington Post reporter Rob Stein’s article on the latest frightening report to come from the Centers for Disease Control and Research. The latest report is that efforts to get teenagers to delay sex and use condoms is “faltering.” But by frightening and faltering, they mean that the data suggest we’ve hit something of a plateau after years of improvements.

Anyway, I come across this paragraph:

Coming on the heels of reports that one in four teenage girls has a sexually transmitted disease and that the teen birth rate has increased for the first time in 15 years, the data is triggering alarm across the ideological spectrum.

I’ll just quote the subhead from the National Journal article we discussed this week:

Does one of every four American teenagers really have a sexually transmitted disease? No, despite headlines given to a recent federal study.

Somehow I think that this one-in-four statistic will prove very hard to kill.

Print Friendly

  • Pastor K

    Kinda like the old “Four out of five dentists recommend Trident for their patients who chew gum.”

    Not once in my life has my dentist asked me if I chew gum, and then recommend Trident. And I think I’ve had more than five dentists in my life, so far….

  • Jerry

    Are we seeing more than the common versus strictly medical definition of “disease” here? Isn’t the real point how many people are having sex before marriage and the medical results of that sexual activity?

  • Mollie

    well, to me the problem is that we’re using a really bad 30% error rate study of 600 girls to make sweeping claims about disease rates.

  • MattK

    31.7% of all statistics are made up.

  • C. Wingate

    Actually, I have to disagree with you Mollie (and, well, you too, MattK). While indeed the dubious “1 in 4″ doesn’t go away, at least the main parts of the article are a more sober analysis focused on trends rather than on raw numbers, modulo reporting the routine posturing by the Designated Factions. I think the latter could have been confronted a little more aggressively; the reporter should have played up the fact that that the study didn’t collect the kind of data needed to buttress claims about the effect of the type of sex ed offered. But the statement from Brown that they don’t know why the trend has changed is placed prominently.

    The Routine Statements, however, add an odd note in another way: in essence, the factions are seeing religious ghosts. The lead sentence is scrupulously neutral, indeed perhaps over-implying that the remedies of both sides aren’t as effective as they once were. The negative consequences are those that most everyone can agree as being negatives; there’s nothing from the reporter that gives even a hint value-laden words like “promiscuity”. Except for the faction fight 2/3s of the way through, it’s very straight public health reporting. That fight, by contrast, gives the impression that sexual license is the only thing that the two factions are concerned about, and that the statistical story is only a peg on which to hang their standard arguments.

  • Mollie

    C. Wingate,

    I didn’t really comment on the article so I’m not sure if you’re disagreeing with me.

    I did think it wasn’t critical enough. Our previous discussion of CDC reports has noted that the agency has an incentive to present information in a way designed to secure funding. This most recent report is no exception.

    But I thought that the statistical presentation and point-counterpoint of the piece wasn’t bad.

  • Dave


    Where do you get that 30% error rate for a study of 600 girls? I would expect such a study to have an error rate around 4%.

  • Mollie


    From the National Journal article I highlighted a few days ago:

    The data’s complexity and the small sample size mean that the study’s “relative standard error” is greater than 30 percent, according to briefing charts presented by Forhan two days after the press conference. Anne Elixhauser, a senior research scientist at the federal Agency for Healthcare Research and Quality, says that her researchers generally reject data that hit the 30 percent RSE level. With such a high error rate, each year’s results could be “unbelievably different,” she said.

  • Chris Bolinger

    To quote Barbie, “Math is hard.” I minored in math, and I still have trouble remembering all of the fun facts related to statistics. I don’t excuse the ridiculously bad reporting of survey results and other statistics, but I understand that getting it right takes effort and that tight deadlines can lead to oversimplifications and outright mistakes.

  • Harris

    I suspect the other item that has so far been missed is the part where the CDC is champion of a particular vaccine. And coincidentally, at 18%, the bulk of the 1-in-4 happens to be that of HPV, the target of the vaccine.

    Rather than look for religious ghosts, there may be the old fashioned commercial one in the wings.

  • Chris Bolinger

    Good call, Harris. Pharmaceuticals are major power brokers in D.C., and the CDC is a federal agency.

  • Dave

    Chris (#9):

    I looked at Mollie’s reference. Apparently the error figure arises from the fact that there was no simple study of 600 individuals but a series of earlier studies of much smaller sample size were batched together. The final report inherits the wide error bars of the earlier studies.

    Math is hard? Math is fun! ;-)

  • Chris Bolinger

    Dave, makes sense, thanks!