Field Ch. 11 - Moderation/Mediation Flashcards
What is a mediator?
- part of the causal pathway of an effect
- tells you how or why an effect takes place
- “IN” the relationship btwn ind and dep variables (med >i< ator)
Recall the example of socioeconomic status, parental education levels, and child reading ability
socioeconomic status affects reading ability via the mediator of parental education levels
What’s the difference between full and partial mediation?
full mediation: no relationship w/o the mediator in the model
partial mediation: mediator only partially explains relationship
What are the 4 conditions of mediation?
- predictor must significantly predict outcome (model 1)
- predictor must significantly predict mediator (model 2a)
- mediator must significantly predict outcome (model 2b)
- predictor must predict outcome stronger in model 1 (predictor –> outcome) than model 2ab (predictor –> mediator –> outcome)
What is a moderator?
- influences the level, direction, or presence of a relationship between variables
- OUTSIDE the relationship between variables (m >O< derator)
What types of variables can be moderators?
- Categorical: ethnicity, race, religion, fave color, health status
- Quantitative: age, weight, height, income
Recall example about work experience, gender, salary
- work experience predicts salary
- gender changes (moderates) that relationship
(work experience doesn’t predict gender, therefore gender is not a mediator)
Which model is mediation in the SPSS process menu?
mediation = Model 4
Which model is moderation in the SPSS process menu?
moderation = Model 1
SPSS moderation step 1
- create interaction term
(transform > compute > var1 x var2) - run linear regression to check significance of mod1, mod2, and R^2 change
- block 1 has var1 and var2
- block 2 has interaction variable
- stats menu: model fit, estimates, R^2 change
SPSS moderation step 2
Run Hayes PROCESS to examine the effect
- Analyze > Regression > PROCESS
- variables into appropriate boxes
- choose MODEL 1
SPSS mediation steps
- Analyze > Regression > PROCESS
- variables in appropriate boxes
- choose MODEL 4