Page 3 - Moderation and Mediation Theory and Concepts Flashcards
What are conditional effects in moderation
The relationship between a predictor and the outcome variable is dependent on the other predictor (moderating predictor (b3XMi) having a specific value (0)
When should we use centering
When a predictor value of 0 is not possible (normally on b2 [moderator]
What is grand mean centring
For each variable, subtracting the mean from each score
Does centering assist the interpretation of ‘lower order’ (conditional) effects?
Yes, it alters the b value.
Particulary when there is no moderation effect
Does centering assist the interpretation of ‘higher order’ (interaction) effects?
Yes, it does not alter the b value. Particulary when there is a moderation effect
How are centring variables and multicollinearity related?
Reduces the strength of the correlation between the predictor(s) and interaction term
How do you perform centring in SPSS
PROCCESS function
*Identify the grand mean from descriptives
The combined effect of two or more predictor variables on an outcome is know as…
Moderation (conceptually) Interaction effection (Statistical term
What does the Johnson-Neyman Technique do?
Same analysis as simple slopes, but with more detail rather than just three levels of the moderator (1SD, 0, -1SD)
*Produces zones of significance instead
What is the Zone of significance
Region in which values of the moderator results in a significant relationship between the predicitor and the outcome
What is the Zone of significance
Region in which values of the moderator results in a significant relationship between the predicitor and the outcome
A mediator is a variable that helps explain…
WHY or HOW a predictor variable relaties to an outcome variable
What are the key objectives od mediation
Quantify what proportion of the relationship between the predictor and outcome is;
A. Direct from the predictor
B. What proportion flows indirectly to the outcome via the mediator
What are the three effect pathways of mediation?
Total effect (c) Indirect effect (a.b) Direct effect (c' or c prime) =
What is the Total effect (c)
The effect of the predictor on the outcome variable
What is the Indirect effect (ab)
The effect of the predictor opn the outcome variable, through the mediator
Direct effect (c’ or c prime)
The effect of the predictor on the outcome variable, controlling for the mediator
What does path ‘a’ represent
The effect of the predictor on the mediator
What does path ‘b’ represent
The effect of the mediator on the outcome, controlling for the predictor
Baron and Kenny (1986) steps to mediation
- Determine predictor significantly predicts the outcome
- Determine that the predictor significantly predicts the mediator
- Determin the presence of the mediator in a regression model with thepredictor and outcome reduces the predictive strength of the predictor in the model
What is the Sobel test
*Added to Baron and Kenny (1986) to test the significance of the indirect effect not provided view Baron and Kenny
When can we state that the mediator has a significant effect of the relationship between the IV and DV in SPSS?
When the Confidence interval for the indirect effect (A x B correlations) does not cross 0
Examples
Significant = [0.09, 0.47]
Non=significant = [-.07, .047]
When can we state that the mediator has a significant effect of the relationship between the IV and DV in theory
If the proportion of variance attributable to the indirect effect (A.B) via the proposed mediator is significant we say we have statistical support for the mediator
Full mediation
If the proportion of variance attributable to the direct effect of the predictor is not significant we say we have support for full mediation
Partial mediation
If the proportion of variance attributable to the direct effect of the predictor is significant we say we have support for partial mediation
Is the Baron and Kenny (1986) method flawed?
Yes, Sobel test of indirect effect is more appropriate
Is causal, when significant mediators do not ensure causation - need solid theoretical base, RCT, longitudinal
What does the Sobel test inappropriately assume?
Assumes normal distribtution for the indirect effect
*unstable results for small samples
What did Preacher and Hayes develop?
Boostrapped test for the indirect effect which does not assume normality, thus avoid Sobel test issue
*Used through PROCESS macros in SPSS
A moderator is a variable that helps explain/answer…
for WHOM and WHEN a relationship occurs between the predictor and the outcome variable
What is moderation?
A technique used to explore whether the relationship between variable changes as a function of a third variable (moderator) [interaction effect
What does centring involve?
The transformation of scores around a designated point that makes interpretation meaningful
