time invariant covariates in LGM

[rothmi...@hotmail.com]

Hey everyone

I`m writting my masterthesis in psychology and I`m struggling with latent growth curve models....

I have a model that fits quite good. No I`d like to add a time-invariant covariate. How can I do this? The covariate is metric (communication scores)

Here`s my model: 

modell_PZ_f <- "
              #first order factors, factor loading equality constraints
              ACSIf =~ a*acsi01.1 + b*acsi05.1 + c*acsi07.1 + d*acsi10.1  
              CCSIf =~ a*ccsi01.1 + b*ccsi05.1 + c*ccsi07.1 + d*ccsi10.1
              HCSIf =~ a*hcsi01.1 + b*hcsi05.1 + c*hcsi07.1 + d*hcsi10.1
              ICSIf =~ a*icsi01.1 + b*icsi05.1 + c*icsi07.1 + d*icsi10.1
             
              #item intercepts, intercepts equality constraints
              acsi01.1 ~ 0*1; ccsi01.1 ~ 0*1; hcsi01.1 ~ 0*1; icsi01.1 ~ 0*1
              acsi05.1 ~ m2*1; ccsi05.1 ~ m2*1; hcsi05.1 ~ m2*1; icsi05.1 ~ m2*1
              acsi07.1 ~ m3*1; ccsi07.1 ~ m3*1; hcsi07.1 ~ m3*1; icsi07.1 ~ m3*1
              acsi10.1 ~ m4*1; ccsi10.1 ~ m4*1; hcsi10.1 ~ m4*1; icsi10.1 ~ m4*1
             
              #item residual variances, no constraints
              acsi01.1 ~~ acsi01.1; ccsi01.1 ~~ ccsi01.1; hcsi01.1 ~~ hcsi01.1; icsi01.1 ~~ icsi01.1
              acsi05.1 ~~ acsi05.1; ccsi05.1 ~~ ccsi05.1; hcsi05.1 ~~ hcsi05.1; icsi05.1 ~~ icsi05.1           
              acsi07.1 ~~ acsi07.1; ccsi07.1 ~~ ccsi07.1; hcsi07.1 ~~ hcsi07.1; icsi07.1 ~~ icsi07.1
              acsi10.1 ~~ acsi10.1; ccsi10.1 ~~ ccsi10.1; hcsi10.1 ~~ hcsi10.1; icsi10.1 ~~ icsi10.1
             
              #item residual covariances, no constraints
              acsi01.1 ~~ ccsi01.1; ccsi01.1 ~~ hcsi01.1; hcsi01.1 ~~ icsi01.1
              acsi05.1 ~~ ccsi05.1; ccsi05.1 ~~ hcsi05.1; hcsi05.1 ~~ icsi05.1
              acsi07.1 ~~ ccsi07.1; ccsi07.1 ~~ hcsi07.1; hcsi07.1 ~~ icsi07.1
              acsi10.1 ~~ ccsi10.1; ccsi10.1 ~~ hcsi10.1; hcsi10.1 ~~ icsi10.1
             
              #first-order factor means
              ACSIf ~ 0*1
              CCSIf ~ 0*1
              HCSIf ~ 0*1
              ICSIf ~ 0*1
             
              #first order factor variances
              ACSIf ~~ vl*ACSIf
              CCSIf ~~ vl*CCSIf
              HCSIf ~~ vl*HCSIf
              ICSIf ~~ vl*ICSIf
             
              #second order growth factors
              i =~ 1*ACSIf + 1*CCSIf + 1*HCSIf + 1*ICSIf
              s =~ 0*ACSIf + 15*CCSIf + 27*HCSIf + 53*ICSIf
             
              #second order factor means and variances (no constraints)
              i ~ 1
              i ~~ i
              s ~ 1
              s ~~ s
             
              #first and second order factors covariation (all fixed to 0)
              ACSIf ~~ 0*CCSIf; ACSIf ~~0*HCSIf; ACSIf ~~0*ICSIf
              CCSIf ~~ 0*HCSIf; CCSIf ~~0*ICSIf
              HCSIf ~~ 0*ICSIf
              i ~~ 0*ACSIf + 0*CCSIf + 0*HCSIf + 0*ICSIf
              s ~~ 0*ACSIf + 0*CCSIf + 0*HCSIf + 0*ICSIf"


modell_PZ_f_fit <- sem(modell_PZ_f, data=mydata, meanstructure=TRUE, missing="fiml", estimator="mlr")
summary(modell_PZ_f_fit, fit.measures=TRUE, standardized=TRUE)

thank you very much in advance!

best

Michelle

[Terrence Jorgensen]

I`d like to add a time-invariant covariate. How can I do this? 

Use the regression operator: ~

http://lavaan.ugent.be/tutorial/syntax1.html

i + s ~ commscores

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