February 28, 2005

On the Importance of E(ui|Xi)=0; Or, Yes, Some Overweight People May Be Systematically Different

By Ian

Many Diet Studies Lack Key Data

Many of the reports omit key mitigating details about the dieters themselves -- things such as medication use, health status, ethnicity and even age.

And later in the artcile:

Gibson's team looked closely at how the articles reported the physical, background and health characteristics of the study participants. As a guide, they used the Consolidation of the Standard of Reporting Trials Characteristics (CONSORT). This is a list of 21 different elements considered essential for a study to be validated by experts and editors of medical journals.

Gibson's team especially focused on age, gender, general health information, use of medication (other than drugs used to control weight), ethnicity and female participant's menopausal status.

Their findings: 92 percent of the studies did not report medication use, while 34 percent ignored the health status of the persons. Ethnicity was not mentioned in 86 percent of the studies; ages were missing in 11 percent.

Eight percent of studies didn't say whether women were pre- or postmenopausal; 4 percent didn't differentiate men from women when reporting results.

That some scientific studies are less than perfect in their methods is no real surprise. And honestly, I've never been one to call people on the carpet for a lack of perfection. I'd be too afraid of all the stories from my own glass house crashing down on me. But I think it's interesting, since such studies directly inform the creation of national policy such as the Dietary Guidelines for Americans. This, of course, does little to change what I'm having for lunch...but it does affect people who are now subject to the whims of legislators who feel the pressure to "do something about this!". (Or, as Helen Lovejoy likes to ask, "Will someone please think of the children?!")

(NOTE: The headline has changed to satisfy my raging pedantry. The use of "may" is more applicable than "might", which I had before. Why I still sometimes let typos go, I cannot explain.)

Posted at February 28, 2005 01:28 PM


You are much too kind in your conclusion.

What is really going on here is junk science. It is not real science.

Comment by Buzzcut at March 1, 2005 10:05 AM | Permalink

Post a Comment

Remember Me?

(you may use HTML tags for style):

Note: You may have to reload to see your comment.

Trackback Pings

TrackBack URL for this entry: