Correlation is not causation

Most people like to have simple explanations to events happening around them. They have a tendency to oversimplify things and give a direct cause and effect explanations to events. But in reality, most events have multiple causes and can rarely be linked to a direct single cause.

Here’s a story to explain this.

Tim attends a party at his friend Steve’s house. At the party, Steve forces Tim to take part in a drinking game. Tim enjoys the game and becomes highly intoxicated. After the party is over, Tim grabs his car keys to go home. Even though Steve knows that Tim is drunk, he does not stop him from driving back home.

On the way home, Tim jumps a red light and smashes into a car. The driver of that car who was not wearing his seatbelt is mildly injured. A passerby looks at the accident but does not offer help to the injured driver. A while later, someone offers help to the injured car driver and takes him to the hospital.

At the hospital, the senior doctor is unavailable that night and so the junior hospital staff takes a while before the driver is admitted and treatment begins. By this time, the initial mild injury is now exaggerated and the patient is now in severe pain which requires urgent attention from the senior doctor. However, since the doctor is not available, the nurse calls him on his phone to brief him about the patients condition. The doctor tells the nurse that he is on his way to the hospital and meanwhile he instructs the nurse to administer certain temporary pain relief drugs to the patient.

The nurse mishears the instructions and administers an overdose to the patient. This dose turns fatal and the patient dies.

Who was the cause?

So, who caused the driver’s death?

  • Was it Steve who pushed Tim to get drunk and did not stop him from driving in an intoxicated state?
  • Was it Tim because he was irresponsible and drove drunk and jumped the red light?
  • Was the dead driver himself at fault for not wearing his seatbelt?
  • Was it the passerby who chose to ignore the accident and did not offer help?
  • Was it the senior doctor who was not present during his duty hours?
  • Was it the nurse who misheard and administered an overdose?

The causes can even be expanded further to non-direct elements like:

  • Was the hospital administration responsible for not implementing better procedures to ensure that patients received the right treatment?
  • Was the car manufacturer responsible for poor crash design?
  • Was the city council responsible for not strictly implementing seatbelt rules?
  • Was the inventor of the drinking game responsible?

This list can go on and on and on to infinity.

In reality, people (and laws) will most likely blame Tim for the death of the driver as Tim was DUI. But when his lawyer takes up his case, he has 10 different ways to interpret and blame the cause of the accident.

Correlation is not causaulity

In the story above, if you remove or change any one of the events from the sequence, the driver would not have been dead and so it’s really difficult to ascertain the event to a single cause.

All those events were correlated to each other but were not the cause. Similarly, we need to stop looking for simple explanations to events around us. In most cases, it’s really difficult to pinpoint a single cause and effect for an event.

We see this happening around us all the time, especially in news headlines:

  • The stock market crashed because FED hiked interest rates
  • Gas prices have increased because Russia attacked Ukraine
  • I fell ill because of winter season

These simple explanations are good enough to make sense in common situations but for being able to take better decisions, we need to be aware of our mind’s tendency to equate correlation to causality. We need to be able to see through an event and dissect the multiple causes it could have and then assign probabilities to all those causes.