Scribd
Upload
178 views2 pages

Assessing Mediators and Moderators

This document outlines the steps for assessing whether a variable acts as a moderator or mediator in a regression analysis. It explains that moderation involves testing for an interaction effect between an independent and moderating variable on the dependent variable. Mediation involves showing that the independent variable affects the mediator variable, which then affects the dependent variable, reducing or eliminating the direct effect of the independent variable. The document provides detailed instructions for conducting the appropriate statistical tests and interpreting the results to determine if moderation or mediation is present.

Uploaded by

shujat_ali7
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
178 views2 pages

Assessing Mediators and Moderators

This document outlines the steps for assessing whether a variable acts as a moderator or mediator in a regression analysis. It explains that moderation involves testing for an interaction effect between an independent and moderating variable on the dependent variable. Mediation involves showing that the independent variable affects the mediator variable, which then affects the dependent variable, reducing or eliminating the direct effect of the independent variable. The document provides detailed instructions for conducting the appropriate statistical tests and interpreting the results to determine if moderation or mediation is present.

Uploaded by

shujat_ali7
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOC, PDF, TXT or read online on Scribd
You are on page 1/ 2

Mediator/moderator worksheet

Steps for Assessing a Moderator Effect

1) From theory / hypotheses, identify your main predictor of interest (IV1), the potential
mediator(IV2), and the outcome variable (DV).
If both IV1 and IV2 are categorical, simply use the factorial ANOVA technique and
test for significant interaction effects.
If at least one of your IVs is continuous, proceed with check-list.

2) Prepare the data set by (a) centering all continuous variables other than the outcome
variable (compute IV – mean score for the IV) and,(b) if you have a categorical mediator,
dummy code it. Then create a new variable for the interaction term (centered predictor
by centered moderator) and, if your moderator was categorical (with more than 2
categories), dummy code the interaction term.

3) Perform a new regression analysis, with centered predictor and centered moderator in
their appropriate order (within your regression model), and then the interaction term ain
a separate block. “Force enter” all variables.
If the interaction term is significant, then there is a moderation effect. If not, there
is no moderation occurring.

4) To interpret the meaning of the moderator, the moderator itself should not contribute
significantly to the regression mode.
If there is no significant relationship between IV2 and DV, then it is clear that IV2
is the moderator.
However, if both IVs are correlated with the DV, then it is mathematically
impossible to determine which IV is actually the moderator, and which IV is the
main predictor. You have to rely on theory instead.
Mediator/moderator worksheet

Steps for Assessing a Mediator Effect

1) Theoretically/conceptually, does your predictor variable of interest precede the potential


mediator, in terms of causality?
If not, then it is difficult to argue that the mediator is “in between” the relationship
between the predictor and the outcome variable.

2) Examine all the bivariate correlations between the predictor variable, the potential
mediator, and the outcome variable. Are these relationships significant?
If no- then there is no mediation effect (though it might be useful to assess for
moderation). If yes, then proceed.

3) In your main regression model, does the predictor explain a significant proportion of the
variance in the outcome variable?
If no- then there is no relationship to mediate. If yes, then you can proceed to
step 4 to see if the mediator “explains away” that relationship.

4) Perform another regression, this time using the mediator as the outcome variable, and
the predictor of interest as your predictor (omit all other variables in your model). Does
it explain a significant proportion of the variance in the mediator?
If no- then it will not function as a mediator. If yes- then proceed to step 5.

5) Enter the mediator into your original regression model, in the same block as the
predictor variable (force entered). Does the previously significant relationship between
your predictor and your outcome variable become reduced, or non-significant?
If yes, then you have a mediation effect. If no, then you do not.

6) Note that, in a complex model where there are other predictors besides your variables of
interest, they should be retained in the regressions that are run in step 3 and step 5.
The key thing to remember is that, in such a situation, the interpretation involves
discussing the relationships “in the presence of [the other variables].”

You might also like