Wk 9 - Moderation w Regression

Question 1

What 3 methods might we use to examine 3-variable relationships?

Answer 1

Simultaneous prediction - multiple direct effects
Mediation - indirect effect
Moderation - direct effect influenced by 3rd variable

Question 2

What does a moderator do? (x2)

Answer 2

Influences pre-existing relationship between 2 variables

Affects strength of an effect

Question 3

What relationships does/should a moderator have to other variables? (x3)

Answer 3

Ideally independent of predictor
May/not be related to the outcome variable
Enhances/attenuates/reverses the predictor-outcome relationship

Question 4

How does the focus of moderation different from mediation? (x2)

Answer 4

Moderation is about direct effect changes

Mediation is indirect effect

Question 5

What questions are we asking in moderation? (x3)

Answer 5

Does the degree of the moderator modify the effect of the predictor on outcome?
*Does the interaction have an effect beyond isolated variables?
Does an effect apply differently to subsets of people?

Question 6

What question is mediation asking? (x1)

Answer 6

What mechanism underlies the predictor-outcome relationship?

Question 7

What kind of variables can be used in moderation analysis with regression? (x2, x2, x2)

Answer 7

Moderator can be continuous or categorical
*Often measured, but manipulated ok
Predictor - continuous/categorical, measured/manipulated
Continuous outcome - linear regression
Categorical outcome - logistic

Question 8

How is the interaction calculated in moderation with regression? (x1)

Answer 8

Product of predictor and moderator scores`

Question 9

How is the interaction term evaluated in moderation with regression? (x3, x2)

Answer 9

Hierarchical regression (2 stage)
    *Add predictor and moderator
    *Introduce interaction term
Test for significant change in R-square
    *Does interaction improve model fit?

Question 10

What does a significant interaction term imply? (x1)

Answer 10

That effect of predictor on outcome is different depending on level of moderator

Question 11

If the interaction term is significant in moderation with regression, our next questions are? (x2)

Answer 11

How does the moderator affect the relationship?

How do the predictor-outcome slopes differ at high vs low levels of the moderator?

Question 12

How do we evaluate the effects of a significant interaction term in moderation with regression? (x2)

Answer 12

Simple slopes analysis:

*Separate regression analyses at +/- 1 SD of moderator

Question 13

What are the 5 steps in conducting moderation multiple regression?

Answer 13

Mean-centre predictor and moderator variables
Compute interaction variable
Conduct hierarchical regression
If significant interaction term - simple slopes
Plot slopes

Question 14

How do we mean-centre scores ion moderated multiple regression? (x1)
With the desired effect being to? (x1)

Answer 14

Mean-centred score = score - mean

Reduce collinearity with interaction term

Question 15

What can we use to automate the complex process of MMR? (x1)

Answer 15

PROCESS macro

Question 16

What are the consideration for developing hypotheses about moderation? (x1, x2, x2)

Answer 16

Must hypothesise a significant interaction in initial MMR
Hyps are theory dependent
*Expect enhance, attenuate, reverse relationship?
Simple slopes can be significant or ns
*Maybe no predictor-outcome relationship at 1 level of moderator

Question 17

What is necessary before we can infer causality in MMR?

Answer 17

Manipulate predictor and moderator in different studies

*ie,not cross-sectional design

Question 18

Is it possible for moderation and mediation to occur simultaneously? (x1)
And… (x1)

Answer 18

Yes

Needs at least 4 variables

Question 19

What is the focus of moderated mediation research? (x1)

Which could address what 3 general hypotheses?

Answer 19

Indirect effect

The strength of the indirect effect is changed by the Moderator
Different indirect effects are engaged at different levels of the Moderator
Indirect effect of predictor on outcome differs for different groups of people (ie high/low on moderator)

Question 20

What is the focus of mediated moderation research? (x1)

Which addresses what general hypothesis? (x1)

Answer 20

Direct effect

Moderation effect is controlled by 4th mediating variable

Question 21

In moderated mediation with a single mediator, what relationships can be influenced by the moderator? (x3)

Answer 21

Predictor-mediator
Mediaotr-outcome
Or both

Question 22

In moderated mediation with a parallel mediators, what does the moderator do? (x3)

Answer 22

Determines which indirect effect is applicable

*ie, High moderator applied to 1 mediator, low to the 2nd

Question 23

Explain mediated moderation (x3)

Answer 23

Moderating effect of 3rd variable controlled by 4th (mediator)
Mediation explains why direct predictor-outcome depends on level of moderator
*ie explains different predictor-outcome relationships for different moderator levels

Question 24

In mediated moderation, explain the relationships between variables (x2)

Answer 24

Predictor-outcome relationship is moderated by 3rd variable

Predictor-moderator interaction (the moderating effect) is associated with/controlled by 4th (mediating) variable

Question 25

What general question is addressed by mediated moderation? (x1)

Answer 25

What mechanism explains why the effect of (predictor) on (outcome) is different at different levels of (moderator)?

Question 26

What 2 general situations might we apply Moderaton analysis to?

Answer 26

ID boundary conditions - where does an effect start/stop

When strength of effect differs across people