Dredging data to "prove" white meat consumption contributes to cancer

The American Journal of Epidemiology
recently published a study that animal rights and vegan/vegetarian activists
are sure to jump on – Dr. Pramil Singh and Dr. Gary Fraser of the
Center for Health Research at Loma Linda University in California claim
their study of 34,000 Seventh Day Adventists shows that white meat consumption
increases the risk of colon cancer.

Should you give up or cut back
on white meat? Certainly not based on this study, which arrives at its
results by shameless data dredging.

What’s data dredging? Suppose
I wanted to prove that vegetarians who attend animal rights protests have
a higher rate of colon cancer than vegetarians who don’t. So I get
several thousand vegetarians to fill out questionnaires detailing how
often they attend such protests.

At first I’m sorely disappointed
by the results – vegetarians who attend animal rights protests don’t
seem to get colon cancer any more often than those who stay away from
such protests. But that’s not the result I’m looking for, so
I need to get fancy with the data. I start looking at the results from
every possible angle and find an interesting trend.

While there is no general increase
in colon cancer from attending animal rights protests, I do find an oddity
that activists who regularly read Humane Society of the United States propaganda but only occasionally attend People for the Ethical Treatment of Animals-sponsored protests have lower rates of colon cancer than those
who rarely read HSUS propaganda but regularly attend PETA protests.

And so I publish my results –
I have proven that attending PETA protests increases the risk of colon
cancer.

Convinced?

This is the sort of methodology
Singh and Fraser use. Their study, in fact, found there was no statistically significant association between read meat consumption and
colon cancer and white meat consumption and colon cancer. But that didn’t
produce the result they wanted, so they went searching for associations
in subgroups of data and found a couple.

Specifically the subgroup in their
study who reported they ate red meat more than once a week but ate white
meat only occasionally had a 90 percent increased risk of colon cancer,
while those who consumed white meat more than once per week but only occasionally
eat read meat had a 229 percent increased risk of colon cancer.

Source:

Meat consumption and colon cancer. American Journal of Epidemiology, October 15, 1998.

Study finds chicken cauces cancer too. Reuters, October, 1998.

Leave a Reply