4B Flashcards
What are the key questions of causal processes?
- What is the direction of the causal process?
- What are the steps of the causal process?
- To whom does the causal process apply?
- On what level does the causal process take place?
Why are theoretical research designs important in statistical analysis?
Some statistical methods (e.g., multiple regression, mediation, moderation) help analyze theoretical problems, even if students never apply them directly.
What is an example of a causal relationship in political science?
The relationship between citizens’ issue attitudes and their left-right identification.
What is a spurious correlation?
A statistical correlation between two variables caused by a third variable (e.g., income and education affecting both issue attitudes and left-right identification).
How can a spurious correlation be tested?
By using a multiple regression model with control variables to see if the relationship still holds after accounting for confounding factors.
What is a reversed causal relationship?
When the assumed cause (X) is actually the effect (Y).
For example, left-right identification might determine issue attitudes instead of the other way around.
What is a bidirectional causal relationship?
When X influences Y, but Y also influences X, creating a two-way causal effect.
How can bidirectional causation be tested?
Using a panel study with lagged effects, which examines whether past values of X predict future values of Y, and vice versa.
What is a causal network?
A system where multiple variables influence each other, rather than a single direct effect from X to Y.
What is a feedback loop in causal networks?
A process where an effect reinforces itself (positive feedback) or counteracts itself (negative feedback).
What is a mediation effect?
When the relationship between X and Y operates through an intermediate variable (Z).
What is an example of mediation in political science?
Education and income may influence issue attitudes, which in turn influence left-right identification.
Issue attitudes act as the mediator.
How can mediation effects be tested?
Using mediation analysis, which assesses whether the effect of X on Y is partially or fully explained by an intermediate variable.
What is moderation in causal analysis?
When the relationship between X and Y depends on a third variable (moderator), which changes the strength or direction of the effect.
What is an example of moderation?
The effect of issue attitudes on left-right identification is stronger among highly educated citizens than among less-educated citizens.
How can moderation effects be tested?
Using multiple regression with interaction effects, which allows the relationship between X and Y to vary depending on a third variable.
What is an ecological fallacy?
Drawing conclusions about individuals based on aggregate-level data.
What is an example of an ecological fallacy?
If wealthier states vote Democratic, it does not mean that wealthier individuals are more likely to vote Democratic.
Individual-level and state-level relationships can differ.
What is an individualistic fallacy?
Drawing conclusions about groups or countries based on individual-level data.
What is an example of an individualistic fallacy?
Highly educated individuals vote more, but this does not mean that countries with more educated citizens have higher voter turnout.
Why is the ecological fallacy more common?
Aggregate-level data is more widely available than individual-level data, leading researchers to make incorrect assumptions about individuals.
How can data from different levels of analysis be studied properly?
Using multilevel models, which analyze data from multiple levels (e.g., individuals and countries) in a single model.
Why is correlation not causation?
Correlation may exist due to:
1. X causing Y
2. Y causing X
3. A third variable (Z) causing both X and Y
4. X and Y influencing each other
5. X and Y being part of a larger causal network.
What statistical methods are used to analyze causality?
Multiple regression, panel studies, mediation analysis, moderation analysis, and multilevel models.
