Have you ever thought about how the number of people who drown in pools each year might relate to the number of Nicolas Cage films released in the same year? Surprisingly, there’s a 66.6% correlation between these two seemingly unrelated things. Or consider the 99.26% correlation between the divorce rate in Maine and the per capita consumption of margarine. Even more bizarre is the 99.79% correlation between spending on science, space, and technology and the number of suicides by hanging, strangulation, and suffocation.
Just because two things are correlated doesn’t mean one causes the other. This is a common logical fallacy. We often see eye-catching headlines like “people who have more sex make the most money,” but these claims can be misleading. For instance, cheese consumption probably isn’t related to how many people die tangled in their bed sheets, but sometimes the connections aren’t so obvious.
Consider the case of menopausal women taking hormone replacement therapy. Studies initially showed that these women had a lower incidence of heart disease, leading doctors to believe the therapy was protective. However, randomized control trials later revealed that hormone replacement therapy actually increased heart disease risk. The original data was reanalyzed, showing that women on the therapy were from higher socio-economic groups with healthier lifestyles, which was the real reason for their lower heart disease risk.
Another example involves children who used nightlights and were more likely to develop myopia. It turned out that myopic parents were more likely to leave a light on for their children, and parental myopia was the actual cause of the children’s condition. This is an example of a lurking variable, where A doesn’t cause B, but a third factor, C, causes both.
It’s like noticing that people with lung cancer often carry lighters and concluding that lighters cause cancer, without realizing that smoking is the real culprit. Scientists strive to avoid these errors in their studies, but the problem worsens when the media sensationalizes coincidental correlations.
One study about chocolate and weight loss was designed to highlight how science reporting can be exaggerated. A science writer with a PhD in microbiology conducted a real clinical trial with three groups: a low-carb diet group, a low-carb diet plus chocolate group, and a regular diet group. After three weeks, the chocolate group lost the most weight. However, the study used only 15 participants and measured 18 outcomes, increasing the chance of finding a statistically significant result.
If peer-reviewed, the study would have faced scrutiny. Instead, the writer submitted it to a journal for a fee, created a fake institution named the Institute of Diet and Health, and sent a press release to media outlets. Soon, “Slim by Chocolate” was front-page news.
Despite these issues, we can’t dismiss correlation entirely. Correlative evidence is crucial in science. Double-blind studies aren’t always possible or ethical, so correlation often provides the best available evidence. When potential causative relationships are systematically explored, correlation can be a powerful tool for understanding cause-and-effect relationships and advancing scientific knowledge.
Thanks to Tyler Vigen for his intriguing charts on these correlations. Check out his website or his book, “Spurious Correlations,” for more fascinating examples, and subscribe for more weekly science videos every Thursday.
Explore the internet to find five examples of spurious correlations. Create a presentation that explains each correlation and why it does not imply causation. Be prepared to present your findings to the class and discuss the potential dangers of misinterpreting these correlations.
Read a scientific article or study that claims a causal relationship. Analyze the study to identify any potential confounding variables or biases. Write a report summarizing your findings and suggest ways the study could be improved to better establish causation.
Participate in a class debate where one side argues for the importance of correlation in scientific research, while the other side highlights the risks of confusing correlation with causation. Use real-world examples to support your arguments and engage in a critical discussion about the role of correlation in science.
Work in groups to design a simple experiment that tests a hypothesis involving correlation and causation. Identify potential confounding variables and describe how you would control for them. Present your experimental design to the class, explaining how it could help establish a causal relationship.
Analyze a recent news article that reports on a scientific study. Evaluate the article for any sensationalism or misinterpretation of the study’s findings. Write a critique of the article, highlighting any misleading claims and suggesting how the reporting could be improved to accurately convey the study’s conclusions.
Here’s a sanitized version of the provided YouTube transcript:
—
Have you ever wondered how the number of people who drown in pools each year relates to the number of Nicolas Cage films released in the same year? It turns out there is a 66.6% correlation between the two. Or consider the fact that there’s a 99.26% correlation between the divorce rate in Maine and the per capita consumption of margarine, or a 99.79% correlation between spending on science, space, and technology and the number of suicides by hanging, strangulation, and suffocation.
Just because there is a correlation between two variables doesn’t mean that one causes the other; this assumption is a logical fallacy. Yet, we are often drawn to headlines like “people who have more sex make the most money.” Unfortunately, this does not always hold true. For example, cheese consumption probably isn’t related to how many people die tangled in their bed sheets, but sometimes the connections aren’t so obvious.
Numerous studies have found that menopausal women taking hormone replacement therapy had a lower-than-average incidence of heart disease, leading doctors to believe that hormone replacement could protect against heart disease. However, when women underwent randomized control trials, they found that hormone replacement therapy actually increased the risk of heart disease. When the original data was reanalyzed, it was discovered that women who took the therapy were from a higher socio-economic group with better diets and exercise regimes, which was the real cause behind the decreased risk of heart disease.
Another case found that children who used nightlights as kids were more likely to develop myopia. However, there is a strong link between parental myopia and the development of child myopia, meaning myopic parents were simply more likely to leave a light on in their child’s bedroom. These are examples of a lurking variable, where A does not cause B, but rather C causes them both.
It’s similar to observing that people with lung cancer often carry lighters in their pockets and concluding that lighters must cause cancer, without realizing that smoking is the confounding variable. Scientists work hard in their studies to avoid these pitfalls, but it gets worse when popular media takes advantage of potentially coincidental correlations.
One study about the chocolate weight loss connection was actually designed to expose how science reporting can be sensationalized. A science writer with a PhD in microbiology ran a real clinical trial where participants were assigned to three groups: a low-carb diet, a low-carb diet plus a 1.5-ounce chocolate bar, and a group that maintained their regular diet. At the end of three weeks, the chocolate group lost the most weight. However, the journalist used a small sample size of 15 participants and measured 18 different outcomes, which increases the likelihood of finding a statistically significant result.
If this study had been peer-reviewed, it would have been scrutinized. Instead, the writer submitted it to a journal for a fee, creating a fake institution named the Institute of Diet and Health. He then sent out a press release to numerous media outlets, and soon, “Slim by Chocolate” was front-page news.
Despite these issues, we can’t dismiss correlation entirely. Correlative evidence is an essential part of science. Double-blind studies are not always possible or ethical, often leaving correlation as the best evidence available. When possible causative relationships are systematically explored, correlation can be a powerful tool for assessing cause-and-effect relationships and advancing science.
A big thanks to Tyler Vigen for providing his charts on these interesting correlations. You can check out his website or his book, “Spurious Correlations,” for more peculiar examples, and subscribe for more weekly science videos every Thursday.
—
This version removes any inappropriate language and maintains the essence of the original content.
Correlation – A mutual relationship or connection between two or more things where one may predict the other, but does not necessarily cause it. – Example sentence: In science, it is important to distinguish between correlation and causation when analyzing data.
Causation – The action of causing something; the relationship between cause and effect. – Example sentence: Establishing causation in scientific studies requires rigorous experimentation and analysis.
Variables – Elements, features, or factors that are liable to vary or change, especially in an experiment or study. – Example sentence: Identifying independent and dependent variables is crucial for designing a valid scientific experiment.
Science – The systematic study of the structure and behavior of the physical and natural world through observation and experiment. – Example sentence: Science relies on empirical evidence and critical thinking to advance our understanding of the universe.
Critical – Involving careful judgment or evaluation, especially in the context of analyzing information or arguments. – Example sentence: Critical thinking skills are essential for evaluating the validity of scientific research findings.
Thinking – The process of considering or reasoning about something, often involving problem-solving and decision-making. – Example sentence: Effective scientific thinking requires the ability to question assumptions and consider alternative explanations.
Evidence – Information or data that supports a conclusion or hypothesis, often gathered through observation or experimentation. – Example sentence: Scientists must gather substantial evidence before drawing conclusions from their research.
Studies – Research projects or investigations conducted to discover or interpret facts, test theories, or develop new applications. – Example sentence: Longitudinal studies are valuable for understanding changes in variables over time.
Risk – The possibility of loss, harm, or other adverse effects, often assessed in scientific research to determine potential outcomes. – Example sentence: Assessing the risk of new technologies is a critical component of responsible scientific research.
Relationships – The ways in which two or more concepts, objects, or people are connected, or the state of being connected. – Example sentence: Understanding the relationships between different variables is key to developing accurate scientific models.
