Unit 12 – Causality and bivariate causal hypotheses
Synonyms dependent/independent:
Independent variable - Dependent variable
Cause - Effect/consequence
X-variable - Y-variable
Exogenous concept - Endogenous concept
Treatment - Observation
Causality and explanation:
Types of questions:
1. Why did YOU become an alcoholic? (Asking for reasons)
- There is asked for a story, parts of this story could be checked
- Can be a starting point for more general ‘theories’
2. Why did THIS PERSON become an alcoholic? (Suppose the person died)
- Answering using existing knowledge about the case.
- Using causal hypotheses to offer an explanation
- Does not improve our general knowledge about causes
3. Why do PEOPLE Become alcoholics? (general)
- Testing a relationship between a cause and consequence .
Three aspects of causality:
1. Correct time order:
- The independent variable (X) precedes the dependent variable (Y) in time
o Example: The effect of alcohol acceptance during childhood and current
drinking behaviour
2. Association or correlation:
- X and Y are correlated
- If there is no correlation, there is no causality.
o Example: If alcohol acceptance during childhood is equal among alcoholics
and non-alcoholics
3. Non spuriousness:
- No third variable accounting for the association
o Example: Testing family circumstances and alcoholic, but not testing the
genes.
Time order problems:
- Measuring both variables at the same time may produce reverse causation.
Example: does a happy childhood make you more happy now?
- Collect data at different points in time to check time order.
Interrupted time series design= testing a variable before treatment and testing a variable
after treatment.
The effect of third variables /non-spuriousness/ modifier variable:
Relationships can be ‘spurious’ or seriously biased because of:
- Cofounding /explanation
- Interaction / specification / modification
Confounding/explanation=
- when a test variable is introduced, the relation between two variables gets weaker or
disappears. This variable explains the dependent and independent variable.
, - There is no relationship between variables, but the third variable makes it look like
there is a relation.
Interaction/ specification /modification=
- When a test variable is introduced, the relationship becomes weaker or disappears
for one of the variables but is maintained or gets stronger for the other variable.
Bivariate associations: positive or negative relations are also called the ‘sign’ of the
relationship.
- A basis for comparison (variables)
Deterministic and probabilistic relations:
Deterministic: if…. happens, then ‘’always’’ (not possible in empirical research)
Probabilistic: if…. happens, then ‘’relatively more/less often’’
Why probabilistic only?
- Measurement error impossible to measure perfectly
- Parsimonious models: simple models, omitted variables leaving weird variables out
because we want the world to be simple.
Testing hypotheses:
- Only if the expected relationship is deterministic, we can reject the expectation with a
single observation
- In the social sciences, all expected relationships are never deterministic. They are
always probabilistic
- This is partly because of the theory: there are always other variables affecting the
dependent variable too.
Testing a causal relationship by studying/testing/finding variation:
- Within (single) case analysis: (under these circumstances….)
- Between (across) case analysis: (compare people)
- Combined within
Synonyms dependent/independent:
Independent variable - Dependent variable
Cause - Effect/consequence
X-variable - Y-variable
Exogenous concept - Endogenous concept
Treatment - Observation
Causality and explanation:
Types of questions:
1. Why did YOU become an alcoholic? (Asking for reasons)
- There is asked for a story, parts of this story could be checked
- Can be a starting point for more general ‘theories’
2. Why did THIS PERSON become an alcoholic? (Suppose the person died)
- Answering using existing knowledge about the case.
- Using causal hypotheses to offer an explanation
- Does not improve our general knowledge about causes
3. Why do PEOPLE Become alcoholics? (general)
- Testing a relationship between a cause and consequence .
Three aspects of causality:
1. Correct time order:
- The independent variable (X) precedes the dependent variable (Y) in time
o Example: The effect of alcohol acceptance during childhood and current
drinking behaviour
2. Association or correlation:
- X and Y are correlated
- If there is no correlation, there is no causality.
o Example: If alcohol acceptance during childhood is equal among alcoholics
and non-alcoholics
3. Non spuriousness:
- No third variable accounting for the association
o Example: Testing family circumstances and alcoholic, but not testing the
genes.
Time order problems:
- Measuring both variables at the same time may produce reverse causation.
Example: does a happy childhood make you more happy now?
- Collect data at different points in time to check time order.
Interrupted time series design= testing a variable before treatment and testing a variable
after treatment.
The effect of third variables /non-spuriousness/ modifier variable:
Relationships can be ‘spurious’ or seriously biased because of:
- Cofounding /explanation
- Interaction / specification / modification
Confounding/explanation=
- when a test variable is introduced, the relation between two variables gets weaker or
disappears. This variable explains the dependent and independent variable.
, - There is no relationship between variables, but the third variable makes it look like
there is a relation.
Interaction/ specification /modification=
- When a test variable is introduced, the relationship becomes weaker or disappears
for one of the variables but is maintained or gets stronger for the other variable.
Bivariate associations: positive or negative relations are also called the ‘sign’ of the
relationship.
- A basis for comparison (variables)
Deterministic and probabilistic relations:
Deterministic: if…. happens, then ‘’always’’ (not possible in empirical research)
Probabilistic: if…. happens, then ‘’relatively more/less often’’
Why probabilistic only?
- Measurement error impossible to measure perfectly
- Parsimonious models: simple models, omitted variables leaving weird variables out
because we want the world to be simple.
Testing hypotheses:
- Only if the expected relationship is deterministic, we can reject the expectation with a
single observation
- In the social sciences, all expected relationships are never deterministic. They are
always probabilistic
- This is partly because of the theory: there are always other variables affecting the
dependent variable too.
Testing a causal relationship by studying/testing/finding variation:
- Within (single) case analysis: (under these circumstances….)
- Between (across) case analysis: (compare people)
- Combined within
