WEEK 2- L4.1 causality Flashcards
necessary conditions for causality by stuart mill
4
- covariance- simultanous change of two variables
- temporal ordering- cause has to change before the affect (needs to initiate change)
- spatial and temporal contiguity- must be possible to justify a plausible link
- nonspurious- presence of a confounding variable should be ruled out (cause and effect are both caused by a confound)
importance of covariance;
3
covariance is necessary, but not sufficient for causality
deterministic relationship
a cause will always produce the same outcome (perfectly predictable)
probabalistic relationship
presence of a cause makes it more likely that an outcome will happen
what are intervening variables?
2
moderator
mediator
what are antecedent variables?
2
confound, control
confound variable example
we may first think that an electoral system influences the number of political parties. but in reality, the cleavage structure of a country may be causing the electoral system in the first place.
mediator
comes betwen the cause and the effect
IV => Mediator => DV
mediator example
liberal noorms at the society level, doesn’t effect the decision against war directly, but is mediated by individual decision makers
moderator
comes between cause and effect but rather moderates the strength of the outcome
- supressor
- reinforcer
- distorter
IV => DV
moderator example dpt
if a country is a member of a military alliance such as nato, this may affect the decision-making process against war
(in this case, it can reinforce the decision)
moderator example negative ads
we want to examine how negative ads influence their candidate preference. in this case, party identification may be a moderator. if someone is a huge trump fan, and you show them a negative ad about trump, party identification may distort (reverse) the effects that the negative ads were supposed to have
counfound
makes the relationship spurious, both are caused by a third factor aka ultimate iv
IV <= Confound => DV
