Hi Witold,

here are my two cents:

Moderated mediation: A moderator variable alters the strength of the relationship between two other variables, i.e. the predictor and the outcome variable. In a mediator model, this relationship may be the indirect effect or mediator effect. For example, it is possible that a predictor does not have a consistent mediator effect across subgroups (e.g. gender, countries, firms). In one group, the mediator effect is large, while in another group, the mediator effect is zero or negative.

Mediated Moderation: A mediator variable transmits the effect of a predictor variable to the outcome variable. If there is a moderator effect such that the effect of the predictor variable depends on the level of the moderator variable, then this effect changes the mediator variable which then affects the criterion or outcome variable.

Examples from David McKinnon (Introduction to Statistical Mediation Analysis, 2008, p. 279):

Moderated mediation: A mediation process differs for males and females. A manipulation may affect social norms equally for both males and females, but social norms only significantly reduce subsequent tobacco use for females.

Mediated moderation: The effect of a prevention program depends on risk-taking propensity (moderator variable), and this interaction changes a mediating variable of social norms, which then affects drug use.

Kind regards,

Karin

the best explanation is to read the attached 2 recently published articles on moderation and mediation. The method section is helpful.

Hi Witold,

here are my two cents:

Moderated mediation: A moderator variable alters the strength of the relationship between two other variables, i.e. the predictor and the outcome variable. In a mediator model, this relationship may be the indirect effect or mediator effect. For example, it is possible that a predictor does not have a consistent mediator effect across subgroups (e.g. gender, countries, firms). In one group, the mediator effect is large, while in another group, the mediator effect is zero or negative.

Mediated Moderation: A mediator variable transmits the effect of a predictor variable to the outcome variable. If there is a moderator effect such that the effect of the predictor variable depends on the level of the moderator variable, then this effect changes the mediator variable which then affects the criterion or outcome variable.

Examples from David McKinnon (Introduction to Statistical Mediation Analysis, 2008, p. 279):

Moderated mediation: A mediation process differs for males and females. A manipulation may affect social norms equally for both males and females, but social norms only significantly reduce subsequent tobacco use for females.

Mediated moderation: The effect of a prevention program depends on risk-taking propensity (moderator variable), and this interaction changes a mediating variable of social norms, which then affects drug use.

Kind regards,

Karin

Check this link: http://davidakenny.net/cm/mediate.htm

and See attached

Best,

A

Hi Karin,

one question with regard to the mediated moderation: My hunch is that this would be transfered to a SEM by corporporating the mediation variable in a model with product term and estimating the effects of X, Z, and the product variable on this mediator, correct? And the mediated moderation hypothesis would be supported when the indirect effect linking the product variable and the outcome via the mediator would be significant?

Best whishes (hope, you're well!)

Holger

Hi Holger,

not quite, because actually you do not link the product variable to the mediator variable. What you do is to estimate conditional effects of the predictor variable on the mediator variable depending on the level of the moderator variable. These conditional effects are transmitted through the mediator variable.

This is sort of tricky, because the product variable is not a variable comparable to the predictor variable, but it is an artificial variable that you do not measure.

Will you be in Amsterdam at the SEM working group meeting?

Greetings,

Karin

#Karin Schermelleh-Engel

Please explain about the SEM working group meeting. Open or closed forum, how often, what's the thematics?

Best

john-kåre

John-Kåre , the meeting is open to all interested researchers.

The SEM working group was founded in 1986, and since then regular meetings or small conferences have been held across Europe (e.g., Germany, The Netherlands, Belgium, and Switzerland). The group has an informal character; there is no affiliation to any research organization. The main goal of the meetings is to exchange and discuss methodological problems and possible solutions in structural equation modeling and related fields. Presentations about ongoing research are explicitly welcome. For more information please have a look at Jost Reinecke's web site:

https://www.uni-bielefeld.de/soz/personen/reinecke/wg_sem/index.html

Here is a link to the next meeting in Amsterdam:

http://cde.uva.nl/content/events/symposia/2018/3/sem-working-group2.html

Kind regards,

Karin

Hi

Moderated mediation tends to occur when a moderator variable interacts with the possible mediator, in this case the value of the indirect effect would alter depending upon the value of the moderator.

In mediated moderation , the mediator intervenes the specified moderating effect, such that the change in the interaction term alters the moderator.

I would suggest you refer the book "Premier on PLS -SEM" by Hair, Hult, Ringle and Sastedt. They provide graphical representations of these special cases.

Good luck

Best

Avinash

Dear Witold ,

I do not want to repeat what Prof. Karin mentioned. I am pretty sure if you study these articles you find more practical examples about these two advanced analysis:

Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated and mediation is moderated. Journal of personality and social psychology, 89(6), 852.

Sojo, V., & Guarino, L. (2011). Mediated moderation or moderated mediation: relationship between length of unemployment, resilience, coping and health. The Spanish journal of psychology, 14(1), 272-281.

Jun Liu Wei Wang Kun-peng Cao, (2011),"Leader political skill and team performance: a moderated mediation model", Nankai Business Review International, Vol. 2 Iss 1 pp. 5 - 22

Edwards, J. R., & Lambert, L. S. (2007). Methods for integrating moderation and mediation: a general analytical framework using moderated path analysis. Psychological methods, 12(1), 1.

Hope this helps.

Best,