latent change score model probably not identified

simone.j...@gmail.com

unread,

Nov 13, 2018, 9:19:56 AM11/13/18

to lavaan

Hi,

I'm estimating a latent change score model with three waves of data. The latent factor has three indicators at each measurement occasion. My hypothesis is that change between time 1 and time 2 affects change between time 2 and time 3.

I specified the following model syntax:

mymodel2 <- '

#measurement model for three waves of data

tau1 =~ phq_1 + phq_2 + phq_3

tau2 =~ A_phq_1 + A_phq_2 + A_phq_3

tau3 =~ E_phq_1 + E_phq_2 + E_phq_3

#allow for correlations of error terms between measurement occasions

phq_1 ~~ A_phq_1

phq_1 ~~ E_phq_1

A_phq_1 ~~ E_phq_1

phq_2 ~~ A_phq_2

phq_2 ~~ E_phq_2

A_phq_2 ~~ E_phq_2

phq_3 ~~ A_phq_3

phq_3 ~~ E_phq_3

A_phq_3 ~~ E_phq_3

#latent change scores

tau2 ~ 1*tau1

changetau12 =~ 1*tau2

tau3 ~ 1*tau2

changetau23 =~ 1*tau2

#regression

changetau23 ~ changetau12

'

fit1 <- sem(model = mymodel2, data = PHQ_C)

when I estimate the model, I get the following warning message:

# In lav_model_vcov(lavmodel = lavmodel, lavsamplestats = lavsamplestats, :

# lavaan WARNING: could not compute standard errors!

# lavaan NOTE: this may be a symptom that the model is not identified.

Some model parameters are still estimated and the output states there are 13 df. I'm guessing there's a local identification problem with the latent change scores, but I can't figure out what's wrong with the model. Could anyone provide guidance?

Any hints will be greatly appreciated!

Best,

Simone

Terrence Jorgensen

unread,

Nov 22, 2018, 8:47:27 AM11/22/18

to lavaan

#latent change scores
tau2 ~ 1*tau1
changetau12 =~ 1*tau2
tau3 ~ 1*tau2
changetau23 =~ 1*tau2

#regression
changetau23 ~ changetau12
'

I think these models require fixing some variances to zero for identification.

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

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