There is no inherent hardwired relationship between correlation and causation. The more data points we can connect, the more sure we can be, but our ability to show correlation is often a matter of statistics and probability rather than a measure of absolute truths.
Does Correlation Imply Causation?
Correlation does not imply causation, but it can indicate it. The more correlating factors between events, the more likely there is a causal relationship.
- Just because there is a correlation between events doesn’t mean there’s a cause and effect relationship between them. Two events can correlate in just about any way without having any causal connection.
- When many events correlate in several different ways, and those correlations all point toward the same conclusion, it increases the probability that there is a cause /effect relationship. In other words, the more correlating events, the more likely it is there is correlation (See Bayes’ Theorem).
- Sometimes we can find no correlation between events, but there is a cause/effect relationship despite this (the inverse of this theory is also true, absence of correlation doesn’t imply a lack of causation). The truth isn’t dependent on us to prove it, and not everything that is true can be proven true.
- Even if there is a lot of correlating factors, it still doesn’t imply there must be causation.
- There are types of relationships which aren’t causal relationships (i.e. non-causal relationships), like the shoe size example below.
- There are some things that are true, but which we can never confirm as true, and we can prove this with mathematics (see Gödel’s theorem).
An example of Correlation not Implying Causation: Reading ability would seem to correlate with shoe sizes in the United States. Children, who are generally emerging readers, wear smaller shoes. There is a correlation between shoe size and reading ability, but it isn’t a casual relationship. Shoe size isn’t the cause; a third factor “age” is the cause.
Another example of Correlation not Implying Causation: The decrease in swashbuckling pirates since the early 1800’s correlates to a global increase in temperature. There is a relationship of correlation between the two, but not a relationship of causation (or is there?).
An example of Correlation Implying Causation: Sticking with the pirate theme, the increase in the spice trade correlates with an increase in pirates, and both correlate with the rise of modern forms of insurance and stocks (see the story of the stock market, insurance, and pirates).
Correlation can be a big red flag, but correlation alone doesn’t inherently imply causality. When we find events correlate, it is a “clue” that hints for us to start looking for additional factors that might indicate there is a correlation/causation relationship.A video explaining how correlation works.
TIP: There are only two true forms of reasoning, deductive (deducing certain truth from other certain truths) and inductive (inducing likelihood of truth by comparing probable truths). Any time we compare correlating factors to find the likelihood of cause and effect relationships, we are essentially practicing induction (and therefore we are dealing with probability). This is why we can make the blanket statement, “correlation doesn’t imply causation, but it can be a big hint.”
What are Correlation and Causation?
Correlation is a statistical measure of relationships between things; causation is an indication of cause and effect.
- What is Correlation? Correlation is a measure of the relationship between variables and the strength of that relationship. Correlation can be used to make predictions or show connections.
- What is Causation? Causation is the “cause” in a cause and effect relationship.
MYTH: “Correlation Never Implies Causation”
Just because correlation doesn’t prove causation does not mean that “correlation can’t imply causation.” Correlation is often a strong indicator of causation.
In science, when we start seeing relationships of correlation, we do further tests to find other correlating factors. Also, sometimes we look for correlation to determine if events have a causal relationship. Correlation and causation are very useful, but their relationship is an indicator of truth rather than an absolute measure of truth.
An example of correlation implying causation: We have tons of data pointing to climate change being real. The data seems to show that there is a correlation between global warming and the burning of oil, coal, and gas. It is very likely that there is a cause and effect relationship. In this case, correlation points to causation, but it doesn’t prove causation… and this is the point, correlation is very important to consider, but it isn’t the same as absolute proof.
Another example of correlation implying causation: The dog is lying on the floor with cookie crumbs on his face and whimpering like his tummy hurts. There is an open bag of cookies with crumbs leading to his mouth. The dog looks at you with guilt. It is likely that correlation implies causation in this example, still, no matter how many correlating factors there are, all arguments of this type are always probable (not certain).
The Absence of Correlation Doesn’t Imply The Absence of Causation
On the flip side of things, the absence of correlation doesn’t imply a lack of causation.
The fact is, truth exists independently of our ability to study it.
Example: Imagine the dog in the example above. Now imagine the dog is in the backyard acting normally with no trace of cookie crumbs on him. However, the cookie bag is still on the floor open and missing cookies. We find no bite marks, no proof the dog acted. Given this, we struggle to correlate the dog’s actions as the cause of the missing cookies. However, what if I told you that I saw your brother-in-law coming over and letting the dog in the house. What if I saw the dog eat the cookies, saw your brother hide all the evidence aside from the bag on the floor, and saw him put the dog outside? Should you only be looking at the dog’s actions? What about your brother-in-law? Is he a missing third factor? Now, what if I told you that I made the story up, and I ate the cookies? Who do you trust? Does the absence of data imply a lack of causality?
Why is the Relationship Between Correlation and Causation Important?
Let’s say you are doing an analysis for a mining company trying to find out where shale deposits (oil) will be or voting in an election, or trying to figure out whether to vaccinate your child or some other vital issue? What if you are trying to figure out the stock market for your 401k investments? You are going to need to apply some principles of logic and statistics in all the above cases.A video explaining the dangers of assuming the false logic “correlation implies causation.”
Every day you are bombarded with talking points, sometimes they are based on emotions, but often they are a summary of an issue boiled down to one cause/effect relationship. Sometimes talking points have no motive, but other times they are a type of marketing meant to elicit a behavior.
Here are some modern headlines that require you to apply logic: Bacon causes cancer; it snowed, so global warming isn’t happening; gun sales are up because of terrorism; video games make kids violent. How can you tell which of the above are true and which are not? First you need to be able to fact-check and spot falsehoods. An excellent way to spot falsehoods is by reminding yourself that correlation doesn’t imply causation, but may indicate it. When you fact-check headlines, you have to look at the source of the headlines, and examine how the facts and the source you are reading are attempting to show correlation and causation.
If your brain isn’t aware of the dangers of false logic like “correlation implying causality” you will have a difficult time sorting truth from lies. People may not sway you with malice, but you may be influenced without knowing it. Only logic and education can protect you from being misled by sparsely checked factoids and other types of half-truths.