Causation
There’s also Causality? So what is causation, and how do we test it then?
Causality is the area of statistics that is commonly misunderstood and misused by people in the mistaken belief that because the data shows a correlation that there is necessarily an underlying causal relationship.
The use of a controlled study is the most effective way of establishing causality between variables. In a controlled study, the sample or population is split in two, with both groups being comparable in almost every way. The two groups then receive different treatments, and the outcomes of each group are assessed.
Causal Factor
One of the things that affects an event.
Causation vs. Implication
This really twisted my head, but i think it’s the difference between Statistics and Boolean Algebra.
When someone says:
“If I flip the switch, the lights turn on.”
there are two possible readings.
Logical/regularity reading:
- whenever switch-flip happens, light-on happens
That is implication-like.
Causal reading:
- flipping the switch is what makes the lights turn on
That is causation.
More thought
I had a lot of trouble agreeing with this passage.
Consider the following bit of reasoning:
Eating ice cream causes public nudity! Statistics show that consumption of ice cream goes up and down in lockstep with arrests for public nudity.
If you ask someone to explain what is wrong with the above argument, if they don’t say that the person has confused a correlation and a cause, they might say something like, “Well, if you stopped everybody from eating ice cream, that wouldn’t do anything to reduce the rates of public nudity. (In fact, it might increase it, since people wouldn’t be able to cool off by eating ice cream, so they might be more inclined to take their clothes off!)”
This response is quite insightful. When someone makes a claim about a correlation, they are making a claim about the way the world actually is—they are claiming that, among the members of a given population, a higher percentage of Xs than non-Xs have property P. We covered this in a previous module. In contrast, when someone makes a causal claim, they are not so much making a claim about the world as it is. Instead, they are making claims about how the world would be if certain changes were made.
So, if someone claims that eating ice cream causes public nudity, they are, among other things, claiming that if nobody ate ice cream, then there would be less public nudity, and if everybody ate ice cream, then there would be more public nudity.
So the passage is saying claiming causation is a really strong claim.
We say, after all, that smoking causes lung cancer, even though not everybody who smokes gets lung cancer. Similarly, we often speak of one thing preventing another, even though it doesn’t do so in all cases (e.g., regular exercise prevents heart disease, use of contraceptives prevents unwanted pregnancy). In these types of cases, we are speaking of negative causal factors.
So the passage is basically saying:
If X causes Y, then changing X should change Y.
not:
X implies Y
or
Y iff X
Speaking in terms of causal factors instead of causes is useful because it makes clear we are allowing that there are other factors which are also relevant to producing/preventing the effect in question. Indeed, the fact that smoking doesn’t always produce lung cancer leads us to investigate questions like: what’s the difference between those people in whom smoking leads to lung cancer and those in whom it does not? That is, there might be some other factors which, together with cigarette smoking, do guarantee lung cancer. On the other hand, there are plenty of people who get lung cancer without ever smoking, and so we also want to investigate what other sorts of things cause lung cancer.
Thoughts
causality is usually a better mental model than strict necessary/sufficient logic.
So instead of asking:
- “Is hard work sufficient for success?”
- “Is hard work necessary for success?”
it is often better to ask:
- “Does hard work increase the probability of success?”
- “Under what conditions does hard work pay off?”
- “What other factors does hard work need to combine with?”
- “What is the bottleneck in this system?”
That is a more causal way of thinking.
A good model is: success = multi-factor system, not single-switch logic