data analysis & hypothesis testing Flashcards
correlation
A relationship where two variables change together.
causation
change in one variable produces a change in another (requires
certain conditions—time order, absence of confounding variables)
null hypothesis
States there is no relationship or no difference between variable
probabilistic relationships
Knowing one variable’s value increases the likelihood (but not certainty) of a
certain value in another variable
criteria for causality
§ Must be a relationship/correlation between the IV and DV
§ IV must be prior to the DV
Must be no confounds (another variable that is responsible for producing the causal relationship)
what does it mean to reject the null hypothesis? why doesn’t this necessarily prove causation?
Rejecting the null” means the data indicate a statistically significant relationship.
- could be other factors could be influencing the relationship (confounding variables)
How does a probabilistic relationship differ from a deterministic relationship?
Deterministic models allow predicting the future state of the system precisely, while probabilistic models estimate the probable future state without providing an exact prediction
what additional conditions must be met for a causal claim?
establish temporal precedence (the cause must come before the effect) and internal validity (ruling out alternative explanations for the observed relationship
Why can it still be
difficult to establish causation in observational studies even w the 3 key criteria?
- difficulty to manipulate/control all variables
- reverse causality/bi-directional influence; does A cause B, or does B cause A?
difference between correlation and causation
Correlation establishes that a relationship exists between two variables, while causation means that one event results in the occurrence of the other event
