5A Flashcards
What is mediation?
Mediation occurs when the effect of X on Y operates through an intermediate variable Z (X → Z → Y).
What is a mediator variable?
A mediator is a variable that explains part of the relationship between an independent variable (X) and a dependent variable (Y).
What are the three criteria for establishing mediation?
- X must have a significant effect on Y (c).
- The effect of X on Y must decrease after controlling for Z (c’).
- The indirect effect (a × b) must be significant in a Sobel test.
What is full mediation?
When the effect of X on Y disappears completely after controlling for Z.
What is partial mediation?
When the effect of X on Y is reduced but still exists after controlling for Z.
What are the four steps needed to test for mediation?
- Estimate the effect of X on Y (c).
- Estimate X and Z’s effects on Y in one model (b and c’).
- Estimate the effect of X on Z (a).
- Conduct a Sobel test to check the significance of the indirect effect (a × b).
How does a Sobel test help in mediation analysis?
It tests whether the indirect effect (a × b) is statistically significant, confirming whether mediation exists.
What is an example of mediation in political science?
Income may influence attitudes on redistribution, which then influence left-right identification.
Why is distinguishing between mediation and spurious correlation difficult?
In cross-sectional data, both mediation and omitted variable bias can produce similar results, and only theory can determine which applies.
What are the limitations of mediation analysis?
- It does not prove causality, as correlation does not imply causation.
- Statistics cannot distinguish between a mediator and an omitted variable.
- Mediators should not be included as control variables unless mediation is explicitly tested.
What is moderation?
Moderation occurs when the effect of X on Y depends on a third variable (moderator).
What are the three types of moderation?
- The moderator determines if there is an effect (e.g., sex determines whether birth control pills affect parenthood).
- The moderator changes the strength of the effect (e.g., education affects how strongly issue attitudes predict left-right identification).
- The moderator changes the direction of the effect (e.g., corruption determines whether education increases or decreases political trust).
How can moderation be tested?
By adding an interaction term (X × Moderator) to the regression model.
How is the interaction variable for moderation created?
The interaction term is calculated as: Interaction = (X – MeanX) × (Moderator – MeanModerator).
What does a significant interaction term in a regression model indicate?
It shows that the effect of X on Y is moderated by the third variable, meaning the strength or direction of the relationship changes depending on the moderator.
Why is visualization helpful in moderation analysis?
Graphs make it easier to understand how the relationship between X and Y changes at different levels of the moderator.
What is an example of moderation in political science?
The relationship between left-right identification and voting for VVD is stronger among highly educated citizens than among low-educated citizens.
What are non-linear effects?
A relationship where the effect of X on Y does not follow a straight line but curves, such as quadratic relationships.
Why can linear regression models be used to estimate non-linear effects?
By adding a transformed (squared) variable of X, a quadratic effect can be tested within a linear regression framework.
How is a quadratic term created for non-linear effects?
The squared variable is created as: Xsquared = X².
What does it mean if the squared term (X²) is significant?
It indicates that the relationship between X and Y is non-linear.
What is the difference between moderation and non-linear effects?
Moderation examines whether the effect of X on Y changes depending on another variable, while non-linear effects examine whether X’s relationship with Y follows a curved pattern.
