Getting Sick with Food and Statistics

From my former lead agency, the CDC . . .

Researchers said that there was a great deal of guesswork in the new estimates. The 48 million figure is near the midpoint of the range of projections in the survey, which said the number of illnesses could be as few as 29 million or as many as 71 million.

This regarding food borne illnesses.  Anywhere between 29 and 71 million.  That’s a lot of between.  The between for deathstimates is much smaller:  3 to 5 thousand.  But, hey, a billionth here and a billionth there and pretty soon you’re talking real risk to paraphrase the late and great Everett Dirksen, meaning that deceiving with numbers is easy.

But, believe me.  The CDC believes it does science.  State of the art.  And it scientifically can assert that 29,000,000 to 71,000,000 cases of food poisoning occur in the US annually.  Or maybe more.  Or less.  But scientific.  Trust us.

Notice the lack of context, or more technically, a base rate, in these numbers.  How do you contextualize annual cases of food poisoning?  How about this.  There are 300 million Americans.  We eat at least 3 times a day.  There are 365 days in a year.  Thus, each year the nation would have 328,500,000,000 chances to get sick from food (300,000,000 X 3 X 365).  Take the high estimate of food poisoning cases (71 million), divide by the number of chances (328 billion) and that gives an annual national rate of illness of 0.0002%.  The rate for death (5 thousand divided by 328 billion) is 0.00000001%!  For a great illustration of Getting Sick with Food and Statistics, check out the Center for Science in the Public Interest (CSPI), and their Food Alerts – no base rates!  Just those numbers that sound big all by themselves, like . . .

I was running a communication unit in the CDC when they started making scientific assertions about mortality from lifestyle (diet and exercise).  It began at 500,000 which was about the same number as deaths from tobacco use.  Pretty damn scary.  Then shortly after that scientific announcement, the CDC made “improvements” in their science, largely due to the echoing laughter from real scientists about that initial estimate.  The CDC proclaimed that the real scientific number is closer to 365,000 annual deaths.  Then, well, maybe 200,000.  The last time I checked they were claiming (PDF) about 100,000 annual deaths.

If we’re lucky and CDC science keeps improving at this rate, within the next decade, no one will die from lifestyle factors!

Joking aside, please consider the persuasion implications of this kind of work.  How can you possibly convince people to change thinking, feeling, or behaving about anything when your science can’t hit the broadside of a barn in daylight with a blast from a double barreled shotgun?

Of course, you can’t.  And the CDC is a creature of Congress which accounts for the wide variability in everything the CDC does or ever will do.

But, what about you?

You have got to control your science whether the science of peer review literature or simply the science of running a business with your head and not your heart.  You must always understand and obey the Rule that

You Cannot Persuade a Falling Apple.

Be careful what you claim or understand as a Falling Apple.  If folks sitting under the apple tree don’t get hit in the head, then you look foolish.