RMDS Exam summary
Unit 12
Three aspects of causality / testing a bivariated hypothesis
1. Correct time order: the independent variable precedes (voorgaan) the dependent variable.
Example: The effect of alcohol acceptance during childhood and drinking behavior.
2. Association or correlation: if there is no association: no causality
Example: If alcoholic acceptance during childhood is equal among alcoholics and non-alcoholics
3. Non spuriousness: a third variable which influences the outcome.
Example: not testing family circumstances and alcoholic, but not testing the genes.
Time order problems: independent variable should precedes (voorgaan) the dependent variable.
1. Problem may occur when measured at the same time.
2. Measuring both variables at the same time may produce reverse causation
How to check time order: Collect data at different points in time.
No spurious relationship: no third variable affecting the relationship
Two effects:
- Explanation/ confounding: wanneer de test variabele wordt geintroduceerd, de relatie tussen de
twee wordt minder of verdwijnt. De test variabele legt de independent en dependent variabele uit.
Example: storks, babies, urbanization
No relationship between the variables, but the third makes it looks like that.
If the original independent variable precedes the third variable in time. (eerst
independent, daarna third)
- Specification/ interaction / modification: wanneer de test variabele wordt geintroduceerd, de
relatie minder wordt of verdwijnt voor 1 van de variabelen, maar het wordt sterker of blijft voor de
andere variabele. De test variable versterkt de relatie
Example: income, holiday spend, willingness
Association or correlation: if there is no association: no causality
Bivariate associations between variables with various levels of measurement.
- Units for comparison (comparative research)
- A basis for comparison (variables)
, Deterministic and probabilistic
- Deterministic: if… then ‘always (not possible in empirical research)
- Probabilistic: if… then “relatively more/less often
o Only because of measurement error: impossible to perfect measure
o Parisimonious models: simple modles, omitted variables → leaving werid variables
out because we want the world to be simple.
Gebruik nooit PROVE bij empirical onderzoek.
Testing hypotheses
- ONLY if the expected relationship is deterministic, we can reject this expectation with a single
observation.
- In the social sciences, all expected relationships NEVER deterministic, they are ALWAYS
probabilistic.
- This is partly because of our theories: there are always other variables affecting the dependent
variable too.
Variation
The ONLY way to test/study a causal relationship is by finding ‘variation’ (in some way). You cannot
test a causal relationship without variation.
Er moet een verschil zijn in independent en dependent = variation om causaal te zijn.
Unit 15
Research design
The way of answering an explanatory (causal) research question in a convincing way.
The logic, not the logistics of answering such a question. More about thinking than organizing
Distinguish between:
- Research design (example: experiment, cross sectional study etc.
- A data collection method (example: a survey, observation)
- The aim or context of research (example: ex-post evaluation)
- The type of data you try to arrive (qualitative or quantitative)
Cross sectional research
O
- All variables of a set of units are measured at the same time
- None of the variables is manipulated differently for a sub-set of units
- You only can check the association
- You can’t check reverse causation and the third variable
Interrupted time series
O O O O X O O O O
(X= random assignment)
- Studying the same units and variables
over time.
- A type of longitudinal research.
- You can check the association and time
order.
The classic experiment
- Classic is the same as interupted time series but also had a control group.
- Excludes the affect of third variables
! Placebo-effect: control group can occur a placebo effect
→ Double- blind experiment: Both don’t know if they are control or experimental.
Unit 12
Three aspects of causality / testing a bivariated hypothesis
1. Correct time order: the independent variable precedes (voorgaan) the dependent variable.
Example: The effect of alcohol acceptance during childhood and drinking behavior.
2. Association or correlation: if there is no association: no causality
Example: If alcoholic acceptance during childhood is equal among alcoholics and non-alcoholics
3. Non spuriousness: a third variable which influences the outcome.
Example: not testing family circumstances and alcoholic, but not testing the genes.
Time order problems: independent variable should precedes (voorgaan) the dependent variable.
1. Problem may occur when measured at the same time.
2. Measuring both variables at the same time may produce reverse causation
How to check time order: Collect data at different points in time.
No spurious relationship: no third variable affecting the relationship
Two effects:
- Explanation/ confounding: wanneer de test variabele wordt geintroduceerd, de relatie tussen de
twee wordt minder of verdwijnt. De test variabele legt de independent en dependent variabele uit.
Example: storks, babies, urbanization
No relationship between the variables, but the third makes it looks like that.
If the original independent variable precedes the third variable in time. (eerst
independent, daarna third)
- Specification/ interaction / modification: wanneer de test variabele wordt geintroduceerd, de
relatie minder wordt of verdwijnt voor 1 van de variabelen, maar het wordt sterker of blijft voor de
andere variabele. De test variable versterkt de relatie
Example: income, holiday spend, willingness
Association or correlation: if there is no association: no causality
Bivariate associations between variables with various levels of measurement.
- Units for comparison (comparative research)
- A basis for comparison (variables)
, Deterministic and probabilistic
- Deterministic: if… then ‘always (not possible in empirical research)
- Probabilistic: if… then “relatively more/less often
o Only because of measurement error: impossible to perfect measure
o Parisimonious models: simple modles, omitted variables → leaving werid variables
out because we want the world to be simple.
Gebruik nooit PROVE bij empirical onderzoek.
Testing hypotheses
- ONLY if the expected relationship is deterministic, we can reject this expectation with a single
observation.
- In the social sciences, all expected relationships NEVER deterministic, they are ALWAYS
probabilistic.
- This is partly because of our theories: there are always other variables affecting the dependent
variable too.
Variation
The ONLY way to test/study a causal relationship is by finding ‘variation’ (in some way). You cannot
test a causal relationship without variation.
Er moet een verschil zijn in independent en dependent = variation om causaal te zijn.
Unit 15
Research design
The way of answering an explanatory (causal) research question in a convincing way.
The logic, not the logistics of answering such a question. More about thinking than organizing
Distinguish between:
- Research design (example: experiment, cross sectional study etc.
- A data collection method (example: a survey, observation)
- The aim or context of research (example: ex-post evaluation)
- The type of data you try to arrive (qualitative or quantitative)
Cross sectional research
O
- All variables of a set of units are measured at the same time
- None of the variables is manipulated differently for a sub-set of units
- You only can check the association
- You can’t check reverse causation and the third variable
Interrupted time series
O O O O X O O O O
(X= random assignment)
- Studying the same units and variables
over time.
- A type of longitudinal research.
- You can check the association and time
order.
The classic experiment
- Classic is the same as interupted time series but also had a control group.
- Excludes the affect of third variables
! Placebo-effect: control group can occur a placebo effect
→ Double- blind experiment: Both don’t know if they are control or experimental.
