Understanding the science: Why reports conflict

Image from Flickr, by RDECOM

You've probably experienced that awkward moment when you're told via the news that something you were told was bad for you actually isn't.  Eggs are a great example of this.  After years of being told that they lead to heart disease, we learned that they're actually pretty good for us.

Why can't the scientists get it right?

Part of the problem comes from where this information comes from.


If you think about science, you generally picture a group of people in white lab coats conducting experiments to come to a conclusion, regardless of whether or not it's what they wanted it to be.  That's part of it, and that's the part that rarely conflicts with itself.

On the other side are epidemiological studies, and that's a whole other ball of wax.

Epidemiological studies look at a variety of factors and try to find connections between them.  For example, they look at what someone eats, and then various conditions they have been treated for and look for patterns to develop.

These studies may find links between various factors, like a link between red meat and heart disease.  They also find a link between eggs and cholesterol in the blood stream.  These are good things, right?

Not really. 

You see, the flaw is that they find patterns that suggest links between certain foods and certain conditions.  However, some scientists and the media jump all over this and accept it as fact.  There's been no controlled experiments, no elimination of outside factors, nothing.  Just a correlation.

Folks, correlation does not equal causation.

As a result, one study may find a link between a certain food and some medical condition, while another finds there's no link.  The reason is, the first study didn't take into account things like genetic factors, lifestyle outside of diet, or any number of other things.

Now, at this point, it might sound like I'm blasting epidemiological studies out of hand.  I'm not.  They really do serve a valid scientific purpose.  These studies may well suggest that should be looked at through controlled experimentation.  If an epidemiological study finds that processed meats may lead to cancer, then that should trigger other scientists conducting experiments to see if it's really true.

The epidemiological study should show us where to look, rather than be accepted as the answer.  Instead, it really just helps us know what the questions are.

So, next time a news report prattles off about how some food causes something bad, keep in mind that unless they say it was an experiment, it was probably an epidemiological study.  I wouldn't get to worried about it.

Sound science will eventually show it to be way off base...maybe <evil grin>.