Paul’s Perambulations

September 12, 2010

How bad (and potentially dangerous) can public science reporting get?

Filed under: General — admin @ 2:12 pm

Headline in current Time Magazine – HEALTH: BOTTOMS UP. NEW STUDY FINDS THAT HEAVY DRINKERS OUTLIVE TEETOTALERS AND MODERATE DRINKERS LIVE ONGEST OF ALL. First, Time published a retraction (apparently online only) that the study does not include any teetotalers (excluded), but that “abstainers” refers to people who said  they were no longer drinking  (20 years ago). Next, when they control for a number of co-variates, the statistical significance is greatly reduced.  Makes you wonder what if they had identified more co-variates. So no-longer-drinking makes you as unhealthy as heavy drinkers. Oops, we can’t consider that a valid scientific conclusion, for as the researches note, this is not experimental data but self-report data. They try to defend self-report validity but oops again, a page earlier they defended not including teetotalers because of problems with self report. The researchers are data-mining work of other researchers and have to deal with whatever was set up previously. There is genuine medical evidence of the dangers of heavy drinking, which this self-report research supports. But there is no science here to support a conclusion that moderate drinking is healthier than not drinking. And considering that heavy drinking demonstrably can cause health problems, moderate drinking could reasonably be suspect. Something is going on, but statistics per se offer no support for the “have a drink.” message of this Time Magazine article, illustrated with a happy and healthy corkscrew caricature hoisting a drink.  

http://onlinelibrary.wiley.com/doi/10.1111/j.1530-0277.2010.01286.x/pdf for original mediocre article which Time writers misread.

2 Comments »

  1. I may need to be direct with a clarification. I’m not inherently knocking all drinking; I’m just knocking bullshit science reporting.

    Comment by admin — September 12, 2010 @ 2:26 pm

  2. Apparently participants were recruited from clinics associated with Stanford University Medical School. Located in the San Francisco (sub)urban area, the recruitment process from various clinics (unspecified) could readily be biased by locale, e.g., what if “abstainers” were primarily of the Muslim faith? In my personal experience, this would not be surprising if Philadelphia (a city of neighborhoods and immigrants) and San Francisco share any similarities as urban entry cities. I can’t specify what this might mean for results of this survey, but it is one example of various unconsidered variables that might present serious confounds for any interpretation of these survey results.

    Comment by admin — September 12, 2010 @ 5:05 pm

RSS feed for comments on this post. TrackBack URL

Leave a comment

Powered by WordPress