Discover How To Discern The Difference Between Causation And Correlation For A Better Understanding Of The World Around You
The Book of Why is a great guide for challenging assumptions about data and finding the true cause and effect relationship between different variables.
Authors Pearl And Mackenzie show how math and statistics can be used to help answer questions such as why you missed a flight connection or got the last seat in the restaurant.
They explain that it’s important to learn the difference between causation and correlation, so you know if certain factors truly affect your results or if you are just seeing patterns without a real cause.
Reading The Book of Why helps us look for more substantial rules, rather than just assuming things about our life based on anecdotal evidence alone.
It gives us insight into debunking basic “truths” in mathematics, from why people once thought smallpox vaccines were worse than catching the disease itself to which biblical figure ran one of the earliest controlled experiments.
By challenging these assumptions we’re better able to understand how data works and can use it to our benefit.
The Causal Revolution: From Ridiculed To Revered In Just Over 100 Years
Some statisticians have been dismissive of the notion of causation, believing it to be scientifically invalid since proof could not be provided.
This is shown by famed English mathematician Karl Pearson in the early 20th century, who downplayed causation as irrelevant.
He liked to highlight spurious correlations to prove his point, and one of his favorite cases was the claim that a nation consuming more chocolate per capita produced more Nobel Prize winners.
He argued that looking for causation was unnecessary here, yet in actual fact it hides an observable causative factor; wealthier nations with greater resources being more likely to produce advanced scientific discoveries.
This sparked criticism from other scientists, and research from Sewall Wright at Harvard University found that causation could actually be represented mathematically.
By creating a diagram of arrows linking causes and outcomes, along with a path diagram signifying this relationship and an algebraic equation depicting data evidence; ’42 percent of a given coat pattern was caused by heredity.’
Sadly Wright’s methods were largely ignored at the time due to the hostile climate towards causation, but thanks to advances today it is now getting the attention it deserves!
We Must Climb The Ladder Of Causation To Truly Understand Data
Data alone can be misleading when causality is neglected.
As an example, let’s take a look at the smallpox vaccine, which was introduced in the eighteenth century and met with some skepticism due to data suggesting that it caused more deaths than smallpox itself.
However, if we isolate the factors of common cause, we can see that this is not the case; out of 1 million children receiving the vaccine, only 99 were reported to have had fatal reactions while 4,000 would have died without its use.
That being said, it is important to evaluate all available data when reaching conclusions; this means considering both gathered points as well as factors such as age or size.
Even something as seemingly unrelated as shoe size and reading ability share a common connection through age – older children tend to have bigger feet and better reading ability on average than younger ones.
All in all, it is necessary to make sure achieve full understanding of a situation before forming an opinion – by climbing The Ladder of Causation proposed by The Book of Why authors you can get a better picture of what’s going on and draw reliable conclusions based on facts rather than assumptions.
Understanding Cause And Effect Is Essential For Ai To Reach Its Full Potential
The first rung of the Ladder of Causation is all about associations and probability.
This means that it deals with making connections between things, whether it be from observations or from data collected.
Think of it this way – an owl tracking its prey looks only at the movements and tries to figure out where the prey will be in the next moment, without thinking about why it’s moving.
Self-driving cars, meanwhile, can only react to instructions programmed into them and cannot come up with potential reactions a pedestrian may have when they hear a car horn.
Similarly, data collection also falls within this first rung as it requires projections based on passive observation.
For instance, if you were to ask a marketing director to find out how likely a toothpaste buyer is also to buy dental floss, they would collect data on both customers and present the question symbolically as P(floss|toothpaste).
At this stage, we cannot determine whether buying toothpaste or floss causes one another – we can only observe what has happened in the past.
However, such understanding of basics associations and probabilities are necessary before reaching conclusions about causation further up the ladder.
The Power Of Controlled Experiments: Applying The Second Rung Of The Ladder Of Causation To Change Our World
In the book of Why, progress up the ladder of causation is described as requiring more than simply watching the world – it demands we take action and intervene.
Unlike the passive first rung, this second step involves actively influencing events.
An example would be when you take a painkiller to relieve your headache.
That’s an intervention intended to stop the pain – and it works!
Or maybe you’re a marketer wanting to measure how changes in price affect floss sales.
These kinds of questions can’t be asked by computers, so humans are essential here.
And one of the most effective ways for testing out effects is through controlled experiments.
This involves setting up essentially identical groups, with one containing an extra variable that is then compared against those who didn’t have it.
The best part? Doing this isn’t anything new – we have reports from stories from The Bible which demonstrate controlled experiments being carried out thousands of years ago (just look at Daniel’s request for a vegetarian diet for himself and three others against another group on King Nebuchadnezzar’s meat-based diet).
Even modern companies like Facebook rely on interventions and experiments to gain insight into their customers; they often mess around with collections on web pages then compare different groups who see different arrangements against each other in order to determine what resonates best with their users.
Understanding The Three Rungs Of The Ladder Of Causality: How To Identify Complicating Factors In Scientific Studies
The third and final rung on the Ladder of Causality is counterfactuals.
This means being able to imagine what could have happened if a different action were taken.
Counterfactuals are often used in climate science and legal proceedings to answer questions such as, “Would we see intense heat waves if carbon dioxide in the atmosphere were at pre-industrial levels?” or “But for the defendant pulling the trigger, would the victim have died?”
Counterfactuals can help us better understand causal relationships, but presents some challenges due to their complexity.
For example, a computer may not be able to determine whether lighting a match was a sufficient cause of a fire or oxygen was instead simply a necessary cause.
Also, counterfactuals pose difficult questions about complicating factors that should be identified when on different rungs of the ladder in scientific studies.
These difficult questions require deep thought and analysis but can provide valuable insight into complex problems.
The Importance Of Controlling Confounders In Controlled Experiments
Controlling for confounders is essential in establishing causality.
Confounders are influencing factors, associated with the second rung of the Ladder of Causality, that can influence both the participants and outcome of experiments.
For example, if a test group is much younger than a control group on average, then age could become a confounding factor and should be taken into account when making comparisons between groups.
Unfortunately, it is often very difficult to eliminate all confounders from an experiment as was seen during the debate around smoking and lung cancer in the 1950s and 60s with skeptics suggesting that a third variable such as genetics could be responsible instead.
To try counteract this bias randomization can be used.
In this case researchers randomly assign participants to control or treatment groups so that neither the participants nor them know who has been allocated which group which is why placebo drugs are given to those in the control groups in medical trials.
However, randomization isn’t always practical or ethical as it wouldn’t be possible or right to tell people to do something like smoke continuously for 30 years just to test if it would cause cancer!
In addition collecting data from participants who have decided to take prescription drugs without intervention could lead to misleading results due biases such people’s affording capabilities having sway their choice in whether they take part or not.
To counter such issues controlled experiments need to be carried out – known as “do-factors” – so that all relevant factors are accounted for when attempting to establish causality.
Finding The Right Mediator Is Key To Understanding Cause And Effect
The importance of correctly identifying a mediator in establishing correct causality cannot be understated.
Establishing why one thing causes another is the key to understanding causality and how it relates to our lives.
Mediators are like the missing piece of the puzzle that help us understand relationships between factors leading to a certain result.
Let’s take, for example, a house fire.
Smoke from the fire triggers an alarm that signals something has gone wrong.
In this case, smoke is the mediator – the variable that allows us to understand why the alarm is triggered when there is a fire.
This sort of understanding was crucial in preventing scurvy centuries ago, when sailors could combat the disease by drinking citrus juice – although they incorrectly assumed it was because of its acidity.
Vitamin C had yet to be discovered, but just having acidic citrus juice meant scurvy had all but disappeared from naval ranks by 1747.
Such mistakes had deadly consequences on Robert Falcon’s Scott’s South Pole Expedition of 1875 however; due to incorrect assumptions about what caused scurvy, only one member of his crew returned alive after others died with signs consistent with scurvy.
If doctors had understood vitamins at that time and brought citrus fruit on board as preventative measure, things could have been different for them!
In short – correctly identifying a mediator can make prevention and finding cures much easier as it helps uncover relationships among variables which otherwise may have not been known before
Causal Diagrams And Mathematical Formulas: Unlocking The Potential Of Ai To Ask “Why?”
When it comes to understanding the relationship between correlation and causation, mathematical formulae can provide invaluable insights.
By drawing up a causal diagram that links all the factors involved (such as a drug, its effect on blood pressure, and lifespan), we can create equations that show the chances of any given outcome happening.
By breaking down the steps logically like this, it’s possible to turn this information into algorithms that computers can understand.
This means that computers would have access to cause-and-effect models which they could use to answer questions such as “What types of planets could sustain life?” or “Is there a cancer-causing gene?”.
By making use of such equations and algorithms, vast leaps forward in science and medicine would become possible.
It is therefore important that we continue to explore this avenue further in order to unlock its full potential.
Wrap Up
The Book of Why is all about understanding causality, and how failing to understand it has stifled scientific progress.
The book provides a blueprint for logical reasoning that can help determine when correlation implies causation.
Furthermore, this method can be programmed into computers so they can answer causal questions and drive rigorous scientific discovery for the years ahead.
All in all, this book is a powerful reminder that mastering causality has been, and will continue to be, key to uncovering the mysteries of our universe.
