Some of you might have seen/heard the hysteria-inducing news released today that bacon and other red meats “can cause cancer.”
I am always skeptical of these types of announcements, even more so since I started taking food classes and statistics. Research studies are so subjective, regardless of how well they are conducted (and if they aren’t conducted well, they’re even worse). I’m sure the good folks at the World Health Organization (WHO) are all much smarter and more educated than I am, but I can’t help thinking that this was kind of a bone-head move. Like, did they need publicity or something?
Here’s an example that I gave my classes all the time when we talked about understanding research.
Say I am petitioning my city council to put a stoplight at this one corner in my neighborhood. I conduct a poll and do some background research. For the poll, I randomly asked 10% of the people in my neighborhood how they felt about the stoplight. Of those 10%, 90% were in favor of the stoplight. As far as the background research goes, I can prove that 100% of all vehicular accidents in my neighborhood in the last 30 years occurred at that corner.
Sounds like I have a fairly solid case, right? City council is so on my side.
Unless, however, someone actually looks into my research and finds the following:
- Only 54 people live in my neighborhood. That means my “sample” of 10% is about 5-6 people. Notice in my summary of my research above that I did not mention if they were young children, young adults, middle aged parents, middle aged single people, or senior citizens. All of those groups might have different biases for or against the issue. Some of the 5-6 people might be my close personal friends. That would definitely make them biased.
- In the last 30 years, there have only been two accidents in my neighborhood. Yes, they both occurred at that corner. But in 30 years, two accidents are not very significant.
- There is a streetlight on that corner – the only one in the whole neighborhood. It’s possible that, driving at night, people come from the dark street into the bright light of the street corner and have a little trouble visually adjusting. That could also be connected to the accidents at that corner.
- The city is putting a new development into previously uninhabited old farm fields about a mile up the road from my neighborhood. Many citizens worry that the extra drivers this will bring to the area will affect traffic. This could also be connected to people’s desire for a streetlight, in which case the accidents are not the cause of wanting one.
There are so many things that affect everything we could possibly be studying, that all we can really prove is correlation between two factors. Causation, on the other hand (that one factor definitively causes the other), is incredibly difficult to prove.
Another example? Male pattern baldness has a very high statistical correlation to cardiac arrests, meaning men who are bald are more likely to have heart attacks and vice versa.
Does this mean that going bald causes heart disease? Or that heart disease causes baldness?
There is something else out there that connects the two. We just don’t know what.
Correlation, not causation.
If you’d like to know the details of the possible flaws and possible truths in WHO’s group of studies, this article is incredibly interesting and chock-full of information. It’s from one of the blogs I read regularly, Breaking Muscle, which is written by a bunch of nutrition experts and personal trainers. It’s a really good break down of the whole issue. Check it out!