Mediation and Moderation Flashcards

1
Q

What is the difference between moderation and effect modification?

A

Same concepts.

Interaction refers to statistical method to test.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What is difference between mediator and confounder?

A

Mediator and confounder both related to exposure and outcomes of interest but mediator IS on causal pathway and confounder IS NOT on causal pathway.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What is major entity in a study?

A

Unit of analysis that is being analyzed in a study.

What you are attempting to study.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What is unit of observation?

A

What you actually study/item you actually observe.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Does the RQ focus on unit of analysis or unit of observation?

A

Unit of observation.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What is the ecological fallacy?

A

Observed characteristic of a group leads to inference about individual members (i.e. incorrectly making conclusions about individuals based on analyses of group data).

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

What is moderation?

A

The effect (strength and direction) of the exposure on an outcome depends on the level of a third variable.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

How do we test moderation statistically?

A

Through an interaction term.

This is the cross-product of the moderator and the exposure.

See if slopes are parallel

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Can we tell if there is moderation by looking at stratified coefficients?

A

No. Need interaction term.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

How do we know if there is interaction?

A

P-value needs to be significant.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is a mediator?

A

Associated with exposure and outcome and on causal pathway. A mediator account for RL between independent and dependent variable. Has direct and indirect effects.

It is a HYPOTHESIS about a causal netowrk.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What are the steps for interaction?

A

1) Center any continous predictors
2) Enter main effects
3) Enter interaction terms
4) Test interaction term
5) If significant, plot

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What are steps for assessing mediator?

A

1) Show causal variable associated with outcome
2) Show causal variable associated with mediator
3) Show mediator affects outcome

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

What is partial mediation?

A

When partially accounts for RL between exposure and outcome.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is full mediation?

A

When RL between exposure and outcome is 0.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

What is suppression?

A

When the path between exposure and outcome is reversed.

17
Q

What is moderated mediation?

A

The relationship between an exposure and outcome via a mediator depends on level of another variable (moderator).