Moderated mediation: conditional IDEs syntax

[Błażej Mroziński]

Greetings,

I'd like to ask you for some guidance in specifying properly a mediated moderation model in lavaan.

Following Hayes PROCESS macro templates i'm setting model 14:

X is experimental condition coded as 0 - 1, V, M and Y are all continues (mean scores of discrete questionnaire items).

I mean centered all but Y, computed a product of M * V and now I'd like to code all the effects.

mod14 <- '
M ~ a*X
Y ~ c*X + b*M + V + MV

indirect:= a * b
direct := c
total := c + (a*b)'

What I'd like to add here is what output from PROCESS macro calls:

• Conditional effects of the focal predictor at values of the moderator(s):
• Conditional effect of focal predictor at values of the moderator
• Conditional indirect effects of X on Y
• Index of moderated mediation

I woud appreciate some help,

Thank you

[Błażej Mroziński]

Following hints found here

I came up with this syntax:

mod14 <- '
M ~ a*X
Y ~ c*X + b*M + b2*V + b*MV

indirect:= a * b
direct := c
total := c + (a*b)

ab3 := a * b3 #index of moderated mediation?

lowV := a * b1 + ab3 * -0.5
highV := a * b1 + ab3 * 0.5
'

Would this be correct?

'

[Terrence Jorgensen]

Would this be correct?

You need to label b1 and b3 correctly in your syntax, and put parentheses around the added terms that make up the simple slope of the b path.

a * (b1 + ab3 * -0.5)

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen

[Aybike Çavdar]

Hi all,
I had similar problems and mailed it to the group. When I look for answer I found your conversation. Different from the question above, I will run model 15 which moderator has an impact on x to y path. My data set name is IntOrnek and the grouping variable is country. Variables are X (predictor), Y(outcome), M (mediator between X-Y), V(moderator between X-Y and M-Y). Path names are as following:

X -> Y  "c1"
X -> M "a"
M -> Y "b1"
V -> Y  "c2"
VM -> Y "b2"
VX -> Y "c3"

According to that, I wrote the following, 

model15 <- ' M ~ a*X
           Y ~ c1*X + b1*M + c2*V + b2*VM + c3*VX
           conditional indirect := a(b1 + b2*V)
           conditional direct := c1 + c3*V
           total :=  a(b1 + b2*V) +  c1 + c3*V
           index.mod.med := a*b2*c3       #Should I add “b2” as well? is the correct calculation multiplication or addiction?
           Vlow := a*b1 + a*b2*c3*-0.5 # If the previous index calculation correct
   Vhigh := a*b1 + a*b2*c3*0.5
   V ~ V.mean * 1
           V ~~ V.var * V
   total.SDbelow :=  direct + indirect
# Ensure country variable is character
   IntOrnek$country <- as.character(IntOrnek$country)'
 # Fit the multigroup model
> Model15Sem <- sem(Model15, data = IntOrnek, estimator = "MLR", missing = "ML", group = "country")

Does it make sense to compare two samples in terms of the model and effect of the moderator? I am not sure if I have written the indirect and conditional effect codes correctly. Since I am new to R, I find it difficult to interpret the examples I find and adapt them to my model. None of the examples have a moderator on the c' path.

I am not sure if it makes sense to define the model this way.

Any contribution would be greatly appreciated.

Thanks in advance. 

21 Şubat 2019 Perşembe tarihinde saat 23:28:10 UTC+3 itibarıyla Terrence Jorgensen şunları yazdı: