PHIL 12 – Lecture 9 – Introducing Causation
Assignment due Monday
Chapters 17 and 18, Lao (on TED)
• Correlations between properties and statistically significant correlations
o Useful in policy making and personal decision making
o Principle scientific interest is usually necessary to infer causes
• What does it mean to say that A is positively correlated with B?
o Proportion of A’s among the B’s is higher among the non-B’s
• What does it mean to say that a correlation is statistically significant?
o .05% likelihood
• Causal Relationship: x causes y
o Inferring a causal relationship between two properties on the basis of just a small
bit of information along is too hasty
▪ Ex: Owning lung cancer does not have a causal relationship with owning
• Smoking causes people to own an ashtray and have lung cancer
▪ Ex: eating olive oil and having smooth skin
• Being rich allows people to have smoother skin and eat more olive
▪ Ex: smoking a lot of pot and poor performance at school
• Family problems can make people smoke a lot of pot and perform
poorly in school
Correlations As A Result of A Common Cause
• Common Cause: something that makes it so that two properties are correlated
o Ex: smoking causes people to own an ashtray and have lung cancer
• There will generally be correlations between properties with a common cause
• Symmetric: if A is positively correlated with B, then B will be positively correlated with A
o Correlation is symmetric
• Correlation does not imply causation
• Causal links are inferred from information on correlation
• Outline the steps needed to show that receiving private music tuition, such as piano or
cello lessons, is correlated with good performance in the math portion of the SAT.
Explain further why such correlations may indicate, though need not imply, that learning
a musical instrument promotes mathematical aptitude.
What Care About Causation?
• Why care about causation in addition to correlation?
o Want to know what is responsible for a phenomenon, perhaps in order to make
▪ Pony riding and admission to elite university • Rich people are more likely to get into elite universities, and also
know how to ride ponies
• When would mere correlations be useful?
o For making predictions
▪ Ex: markets for ashtrays/stables
▪ Information on correlations is only useful in this sense if there is
something like a common cause relationship
• Purely accidental correlations are not really useful
o Ex: the number of pirates per capita and the global temperature
▪ The proportion of pirates to the general population has gone down as the
global temperature goes up
Causation: Some Controversies
• Thinking about causes raises many deep-rooted philosophical questions
o Does everything have a cause? Is everything an effect of something, or are some
o Are all abuses deterministic, or is there stochastic/probabilistic causation as well?
▪ Ex: smoking and lung cancer
• Not all people