Association vs Cause-and-Effect

You’re being lied to on a regular basis by people who report on science. Sure, many of the people who are guilty of this don't actually realize that they're providing bad interpretations. They themselves have been deceived because they don't know what I'm going to talk about in this post (as well as other errors, which I may discuss at another time). That is, the inappropriate use of associations to infer a cause-and-effect relationship between two factors. I will discuss some specific examples of this type of deceit later in this post, but first, let me tell you a story.
A long, long time ago, in a state far away, on the very first day of my psychology 101 class, my teacher showed us a graph of ice cream sales vs violent crime. As it turns out, there’s a correlation between the two factors. My classmates and myself were then asked to explain this correlation. We, oblivious to how our teacher was fooling us, fell into her trap. Various people tried to explain how eating ice cream makes people more violent, or that violence makes people want ice cream. Both of these conclusions are quite ridiculous. According to my teacher, the reason for the association is that there is a third factor that drives both phenomenon: summer. You see, when it’s hot outside people buy more ice cream, but people are also more irritable and prone to violence. As a result, you can measure as association between ice cream sales and violent crime. After informing us of how she tricked us, my teacher asked us to chant, “correlation is not causation”. (In case you’re wondering, “causation” means exactly the same thing as “cause”, but it makes intellectuals feel good about themselves to use technical sounding words.)
The concept that correlation is not causation is an incredibly important one for anyone who wants to understand science. There are various reasons why you can’t determine cause-and-effect from associations. The biggest is because of the myriads of possible variables that could influence the association. I mentioned a third factor earlier, but there could literally be thousands of factors that may play into a single association.

Definitions

• Correlation: a mutual relationship or connection between two or more things.
• Association: the fact of occurring with something else; co-occurrence

While technically different, the two terms are often used interchangeably, and are used to mean that there is a relationship between two factors. The nature of the relationship can be “positive” or “inverse”. A positive correlation means that as one factor goes up, so does the other factor. An inverse correlation means that as one factor goes up, the other factor goes down.

Strength of an association
The strength of the association is important. An association is said to be strong if the two factors track very closely to one another. An association is said to be weak is the two factors only vaguely track with each other. Two factors are not associated if they cannot track in relationship to each other. One way that the strength of an association is determined is based on the “relative risk”, which is a calculation done to determine the ratio between an exposure and an outcome. Relative risk is expressed as a number. If the relative risk is 1.0, then there is no increase or decrease in the outcome. If the relative risk is less than 1.0, there is a decrease in the outcome. If the relative risk is more than 1.0, there is an increase in the outcome. Keep in mind that “An increased risk of less than 50% (RR=1.0–1.5) or a decreased risk of less than 30% (RR=0.7–1.0) is considered by many epidemiologists to be either a weak association or no association”. (1)
What does the strength of an association mean? If two factors are strongly associated, there might be a cause-and-effect relationship between the two factors. It’s not definitive, but it’s possible. If the association is weak, there is not likely a cause-and-effect relationship, though it’s still possible. If there is no association, then there is no cause-and-effect relationship.

Examples:
Now that we’re done with definitions, let get down to some examples. I’m going to start with the big dogs: The World Health Organization (WHO). They listed “processed meats” as a group 1 carcinogen, meaning that, according to them, there is sufficient evidence that processed meats cause cancer. But what was this classification based on? From WHO’s website “The consumption of processed meat was associated with small increases in the risk of cancer in the studies reviewed” (emphasis added) and “In the case of processed meat, this classification is based on sufficient evidence from epidemiological studies that eating processed meat causes colorectal cancer” (emphasis added, 2) What’s the significance of the statement that it’s “epidemiological studies”? Epidemiological studies are one of the types of studies used to find associations. They do not control variables, they are not an experiment. To be fair, WHO did also consider animal studies and studies exploring possible mechanisms, but they also seem to think that a consistent association supports a cause-and-effect relationship (3). Of course, there’s the strength of the association. Looking through the paper put out by IARC (a branch of WHO), I found 54 different references to relative risk in the studies they included in their analysis related to processed meat and colorectal cancer (the kind of cancer they claim is caused by processed meats). The relative risk ranged in these papers from 0.85-2.0. Only 8 out of the 54 relative risk values was outside of the 0.7-1.5 range, meaning that most of these represent a finding of a weak or absent association. To put this in context, according to the CDC, smoking is associated with a 15-30 fold increased risk of developing cancer (in other words a relative risk of 15.0-30.0) (4). The references I found for red meat, which WHO has labeled a probable carcinogen, and colorectal cancer ranged from 0.8-1.99, with only 2 out of 24 references being outside the 0.7-1.5 range. It seems to me, that WHO has made an inappropriate conclusion from the data they’ve analyzed.
The Nutritional Recommendation Consortium did a more honest analysis. They said, “Our weak recommendation that people continue their current meat consumption highlights both the uncertainty associated with possible harmful effects and the very small magnitude of effect, even if the best estimates represent true causation, which we believe to be implausible.” (5)
For a quicker and lighter example: WebMD.com has an article from several years back, titled “The Link Between Chocolate and the Nobel Prize”, which starts out with “Eat chocolate, win a Nobel Prize? It may sound far-fetched, but a new study suggests it might not be bad advice.” (6) They go on to explain the link (another word for an association) between the chocolate consumption of a nation, and the number of Nobel Prize winners in that nation. After reading this article, hopefully you aren’t thinking that eating chocolate will make you more intelligent. There may be many reasons for this association, and the fact that there are many possible explanations should immediately make you say “correlation is not causation!”
Conclusion
I’ve considered giving other examples, but if you pay attention to science topics (especially those related to health) you will frequently see examples of people touting the newest association to be found, and trying to use that information to convince you that meat causes cancer, chocolate makes you smarter, or that eating ice cream will make you want to attack people, among many other claims. Don’t fall for it.
I hope after reading this post that every time someone brings up an association, correlation or link that you will have an immediate gut reaction of my voice yelling in your head: correlation is not causation! You will then pause, think, and ask a few questions. There may be a cause-and-effect relationship between the two factors, or there may not be, but don’t get pulled in by the sophisticated sounding language, which ultimately just means, “we need more research on this topic before we can draw any conclusions”.
Thanks for reading,
​John Hogue, ND
 
References:

1. Craun, G., Calderon, R., 9. How to Interpret Epidemiological Associations, Gunther F. Craun & Associates Staunton, Virginia, USA, U.S. Environmental Protection Agency National Health and Environmental Effects Research Laboratory Research Triangle Park, NC, USA, Microsoft PowerPoint - WSH05.09Nutrientcover.ppt (who.int) (accessed on 4/6/2021)
2. World Health Organization (2015, Oct 26). Cancer: carcinogenicity of the consumption of processed meat and red meat. Cancer: Carcinogenicity of the consumption of red meat and processed meat (who.int) (accessed on 4/6/2021).
3. International Agency for Research on Cancer (2018, Mar). Red Meat and Processed Meat: IARC Monographs on the Evaluation of Carcinogenic Risks to Humans. IARC Publications Website - Red Meat and Processed Meat
4. Centers for Disease Control and Prevention (2020, Sept 22). What are the Risk Factors for Lung Cancer? CDC. What Are the Risk Factors for Lung Cancer? | CDC (Accessed 4/6/2021)
5. Johnston, B., et al., 2019. Unprocessed Red Meat and Processed Meat Consumption: Dietary Guideline Recommendations from the Nutritional Recommendations (NutriRECS) Consortium. Annals of Internal Medicine. 171(10): 756-764. Unprocessed Red Meat and Processed Meat Consumption: Dietary Guideline Recommendations From the Nutritional Recommendations (NutriRECS) Consortium | Annals of Internal Medicine (acpjournals.org)
6. Warner, J. (2012, Oct 9). WebMD.com.  https://www.webmd.com/food-recipes/news/20121010/link-chocolate-nobel-prize (accessed on 4/6/2021).