View Question
 Question
 Subject: What is causality? Category: Reference, Education and News > Education Asked by: bren-ga List Price: \$10.00 Posted: 12 Oct 2002 20:16 PDT Expires: 11 Nov 2002 19:16 PST Question ID: 75914
 ```If I had to establish causality I understand it is difficulit, whether conclusion has been derived inductively or deductively. Please explain and elaborate on the implication of my statement. Why is ascribing causality more difficult when conclusion have been reached through induction. Correlation does not imply causation. Illustrate this point with business examples.```
 Subject: Re: What is causality? Answered By: haversian-ga on 14 Oct 2002 01:52 PDT
 ```Hello bren-ga, Establishing causality is difficult in the same way that proving something in mathematics is difficult. One must not only show that both events or conditions occur in conjunction, but must prove that the presence of one event or condition always causes the other. In many cases, some unknown third condition causes both the first two conditions. Consider for a moment the electric lightbulb. Electricity passed through a coil causes heat; heat causes the filament to glow and produce light. One might be tempted to say that the same is true of gas discharge lamps which produce heat and light just as incandescents do. In point of fact however, the electricity passing through a gas discharge lamp produces both the heat and light so the two occur simultaneously, but the heat does not cause the light, nor does the light cause the heat. In the mathematical sense, induction is sound. This means that conclusions based on induction upon true things are provably true. In common usage though, induction refers to the process of making a generalization based on incomplete evidence. For example, consider a child learning the meaning of "chair". A child sees a highchair and is told this is a chair. He sees a 4-legged barstool and is told this is a chair. He sees a three-legged stool and is told this too is a chair. The child assumes (induces) that chairs are flat or gently curved seats affixed to legs for support. Then he sees a recliner and is told this too is a chair - the child must now revise his concept of a chair in light of new evidence. Similarly, it may be obvious that causation is present but new information can reveal this to be mere coincidence. I briefly mentioned correlation not implying causation, but you asked for a business example specifically. Studies show that certain reforms precede increased employee morale in 4 out of 5 cases. An ex-CEO returns to a company and institutes these reforms. Shortly thereafter, surveys of employees indicate that morale has increased sharply. Certainly there is correlation, but can we say that the reforms have caused the increase in morale? Not so; the CEO is a popular and dynamic figure whose firing 5 years ago was the subject of much controversy. His return caused the increase in morale rather than his reforms. If I have been unclear, or if you wanted a real-world rather than hypothetical example, or are otherwise unsatisfied, please do not hesitate to request a clarification. -Haversian```
 ```I'm not sure how I'd go with business examples, so I'll just post my thoughts as a comment. Causality requires logical proof that one thing causes another, rather than just a correlation between one thing occuring and another thing occuring. Example: A man thinks that by throwing his newspaper out the window of the train each morning he's keeping his city free from tigers. When someone questions him on this, he says 'but there are no tigers!'. This is inductive reasoning - I do something with the intention of making tigers go away, and there are no tigers, therefore it works. As you can see, the problem is that there may be another event which is keeping the tigers away (say, a big fence and men with guns), or there may not be tigers to keep away at all! Deductive reasoning would go like this (assuming all the premises to the argument of true, which is not the case here but still...) 1. There are tigers in existence in the region of our city 2. The only way to get rid of them is by throwing newspaper out of the train window 3. I - and only 1 - throw newspaper out the train window 4. The tigers have gone since I started throwing newspaper out the train window. Conclusion: I caused the tigers to go away by throwing newspaper out of the train window.```
 ```A good book on Causality is Causality: Models, reasoning, and inference. Judea Pearl. Cambridge University Press, 2000. It discusses several paradoxes and so forth about causality, and develops a Bayesian network based approach to the theory. It turns out that causality is an idea whose depth and complexity is grossly underestimated by most people.```
 ```I will second the comments on the Pearl book. The book is heavy going in places, but fascinating. Pearl's home page is here: http://bayes.cs.ucla.edu/jp_home.html It has some links to some more information, and some talks that he has given.```