The problem with things that “cause” cancer

Causation is a tough concept to wrap our heads around. In its simplest sense, we say that one thing causes another when the first made that second thing happen. This is usually a 1:1 relationship. A leads to B, regardless of whether some other things do or don’t happen, and without A, B would not happen.

One common error is to attribute causality when there is none. It’s this type of thinking that leads us to believe that we need a lucky pencil to take tests – with it, we’ll ace the test; without it, we’ll bomb. When two things are correlated (for example, losing fifteen pounds and getting asked on more dates), it’s easy to make a causal inference, even when it’s not warranted. This is the reason that science teachers drill the phrase CORRELATION IS NOT CAUSATION into students’ heads.

https://xkcd.com/552/

Image: xkcd

We can also make the reverse inferential mistake; that is, when one thing does actually cause another, we can interpret it as a correlation. This is especially true when ascribing to causation would require that we change our behavior. For example, we might be less likely to really buy into the idea that obesity leads to heart disease if it suggests that we should change our habits, instead diluting the relationship to a more correlational one in our minds, acknowledging that, yeah, people who are obese tend to have more heart disease, but there are plenty of obese people who don’t, so maybe there’s no need to cut out the Big Macs just yet. This is commonly referred to as cognitive dissonance: having inconsistent thoughts, beliefs, or attitudes, especially as relating to behavioral decisions and attitude change.

To further complicate causal thinking, many things don’t have 1:1 causes. A might cause B, but only in the presence of C, D, and E, or only in the absence of F and G. And sometimes one of those factors that mediates whether A causes B is pure randomness. This is another concept that is really difficult for humans to wrap our heads around, but randomness has played a huge role in making us the creatures we are and making the world the place it is today.

This week the World Health Organization (WHO) made a splash by releasing guidelines that placed processed meats in the same “cause” category for cancer as smoking and asbestos. What does this mean? It means that the WHO is confident that processed meats increases our likelihood of developing cancer. It does not mean that they increase our chances of getting cancer as much as asbestos or smoking do, but that they are equally confident that all of these things do in fact increase cancer risk. This is not one of those straightforward A causes B types of causation, though. We know that there are some people who eat lots of processed meats and never develop cancer. The causation is one of the more complicated types, most notably involving randomness. If someone eats a lot of these meats and then the right randomness (genetic mutations) take place, that person is more likely to end up with cancer than someone who didn’t eat any processed meat but experienced the same randomness (though that second person could very well get the disease too, as we know).

So the word “cause” is not a lie, or even an exaggeration. It’s true. But how do we interpret it? This week, it seems that most people interpreted it as the 1:1 relationship cause, accounting for much of the media hype. It might seem, then, that we should avoid this chaos-inducing word, and instead go for something less anxiety-provoking: maybe “linked to” or “associated with” would get the job done.

These weaker phrases have their own drawbacks, though, precisely because they induce less alarm. They are likely to encourage more cognitive dissonance, more of the reasoning that this is not something that affects me personally and I therefore shouldn’t feel as compelled to overhaul my sausage-filled diet.

There is probably no single verb that can be used in a headline to capture the relationship between certain behaviors and cancer risk, one that will encourage the right amount of alarm. Our best bet is to be aware that there are no perfect words to talk about complex ideas, and that means we will inevitably use imperfect words, words that mislead in different ways. Sometimes it takes some media chaos for an issue to get the attention it needs so that people can understand a situation and make informed decisions. Hopefully this is one of those times.

 

PS: There is a very cool study of science blogs and blog readers going on! I’ll also be receiving information about survey results from my blog readers, so your responses will be helpful to me as well as the researchers looking to learn more about science blogging more generally. To participate, take this survey: http://lsu.qualtrics.com/jfe/form/SV_0dIyegEdCzOFNxr

For completing the survey, readers will be entered into a drawing for a $50.00 Amazon gift card and other prizes, and all participants will receive a small thank-you gift

Advertisements

2 comments

  1. Interesting post! I liked the point you made here: “We can also make the reverse inferential mistake; that is, when one thing does actually cause another, we can interpret it as a correlation.”

    I think it’s a strong point; we are prone to the fallacy that correlation implies causation… and just as prone to the fallacy that correlation implies a lack of causation. It’s an interesting topic that I enjoyed your exploration on.

    My only suggestion is to try and capture the idea in one sentence that really captures the idea in very plain language. I felt the power of the concept was a bit muted in the sentence above, and I’d love to see it just jump off the page (ok, screen 🙂 ) just a bit more … emphatically 🙂

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s