Plotting interactions in path analysis

[모르모트군]

Hello,

I am trying to plot the interaction term in my path model.

My model:

path1 <-
  'var1 ~ A*zindex+E*zah+
c1*zage+c2*zsex+ c3*zedu+c31*zincome+
c32*var2
DV~ I*var1:var2+
F*var1+G*var2+
c4*zage+c5*zsex+c6*zedu+c9*zincome+c10*zindex'

fit1 <-lavaan::sem(path1, data=mydata, estimator="ml")

I would like to plot the term (var1:var2), like the attached figure, to see the details.

(The figure is not generated by me; it is just an example that I downloaded on google).

Note that intProd() or probe2WayMC() will not work here, since I am conducting interactions among observed variables, not latent variables.

Thank you,

[Terrence Jorgensen]

intProd() or probe2WayMC() will not work here, since I am conducting interactions among observed variables, not latent variables.

One way to hack this would be to treat each observed variable (including the product term, but you should add that as a variable in your data) as single-indicator factors.  Then you should be able to use semTools::plotProbe() 

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

[Kevin Teoh]

Dear Dr Jorgensen, Dear Community, 

I too am trying add an interaction term to a path model, and am trying to work with this hack of treating the IV, DV, Moderator, and Interaction variables as single indicator factors. However, probe2WayMC is bringing up "Error in varEst[targetcol, targetcol] : subscript out of bounds". 

Would anyone have any advise on what I might have missed? I have also tried a basic model without any additional variables/controls; where all the variables are latent factors with a single observed item; and where all variables are observed variables. I get pretty much the same model fit and estimates. 

Any help with the error message is appreciated. I have added script below. 

Thank you

Kevin

# Model estimation with latent interaction

int_model <- '
#loadings
iv =~ IVc
mod =~ MODc
int =~ ICMODdc #doublecentered
dv =~ DV

#regression
mod ~ CTRL1 + CNTRL2 + CNTRL3 + VAR1 + VAR2
dv ~ iv + mod + int + CTRL1 + CNTRL2 + CNTRL3 + VAR3 + VAR4

#covariances
VAR3 ~~ iv + mod + VAR4
VAR4 ~~ iv + mod

'

int_fit <- sem(int_model, data = data, estimator="MLM")
summary(int_fit, fit.measures = TRUE, standardized=TRUE)

# Simple Slopes
probe <- probe2WayMC(int_fit, c("iv", "mod", "int"), "dv", "mod",
c(-sqrt(0.459),0, sqrt(0.459))) #estimators from moderator
probe

plotProbe(probe, c(-3,3))

[Terrence Jorgensen]

"Error in varEst[targetcol, targetcol] : subscript out of bounds". 

I recognize that from a while ago, probably fixed already.  Does this happen with the latest development version?

 https://github.com/simsem/semTools/wiki

Terrence D. Jorgensen    (he, him, his)