one cross panel model

[Sara]

 Hello everyone, 

I am trying to replicate the one cross panel model specified in the article Illegal Drug Use Depressive Symptoms and General Health Exploring Co occurrence across 11 Years in a National Sample by Silver Kelsay and Lonergan 2022 Journal of Psychoactive Drugs DOI 10.1080/02791072.2022.2053003 using the same NLSY97 data but with some differences.

I am working with 5165 observations and have applied imputation for missing data.

Illegal drug use is treated as dichotomous 0 1.  Depression is calculated as the mean of the five MHI 5 items and General health is a single item with six possible responses and is treated as ordinal. 

In the article health and illegal drug use are considered for all years from 2000 to 2010 while depression is measured from 2000 to 2010 but biennially. 

In my study,  I have chosen to consider biennial measurements for all three variables 2000 2002 2004 2006 2008 2010.

Following the model specification in the article, I attempted to estimate it in lavaan using DWLS with parameterization theta and specifying the ordinal variables in the ordered argument. 

However I always receive an error saying the model is not identified.  I am wondering if having more time points as in the paper might help with model identification. According to you how could I simplify the model without losing the substantive structure and meaning of the model.  Are there residual covariances that I could remove to improve identification?

Thanks in advance

  model_clpm <- '

  drug_2002 ~ depression_2000 + health_2000 + drug_2000
  drug_2004 ~ depression_2002 + health_2002 + drug_2002
  drug_2006 ~ depression_2004 + health_2004 + drug_2004
  drug_2008 ~ depression_2006 + health_2006 + drug_2006
  drug_2010 ~ depression_2008 + health_2008 + drug_2008
 
  depression_2002 ~ drug_2000 + health_2000 + depression_2000
  depression_2004 ~ drug_2002 + health_2002 + depression_2002
  depression_2006 ~ drug_2004 + health_2004 + depression_2004
  depression_2008 ~ drug_2006 + health_2006 + depression_2006
  depression_2010 ~ drug_2008 + health_2008 + depression_2008
 
  health_2002 ~ drug_2000 + depression_2000 + health_2000
  health_2004 ~ drug_2002 + depression_2002 + health_2002
  health_2006 ~ drug_2004 + depression_2004 + health_2004
  health_2008 ~ drug_2006 + depression_2006 + health_2006
  health_2010 ~ drug_2008 + depression_2008 + health_2008
 
  drug_2002 ~~ drug_2000
  drug_2004 ~~ drug_2002
  drug_2006 ~~ drug_2004
  drug_2008 ~~ drug_2006
  drug_2010 ~~ drug_2008

  depression_2002 ~~ depression_2000
  depression_2004 ~~ depression_2002
  depression_2006 ~~ depression_2004
  depression_2008 ~~ depression_2006
  depression_2010 ~~ depression_2008

  health_2002 ~~ health_2000
  health_2004 ~~ health_2002
  health_2006 ~~ health_2004
  health_2008 ~~ health_2006
  health_2010 ~~ health_2008
 
  drug_2000 ~~  health_2000
  drug_2002 ~~  health_2002
  drug_2004 ~~  health_2004
  drug_2006 ~~  health_2006
  drug_2008 ~~  health_2008
 
  drug_2000 ~~ depression_2000
  drug_2002 ~~ depression_2002
  drug_2004 ~~ depression_2004
  drug_2006 ~~ depression_2006
  drug_2008 ~~ depression_2008
  drug_2010 ~~ depression_2010
 
  depression_2000 ~~ health_2000
  depression_2002 ~~ health_2002
  depression_2004 ~~ health_2004
  depression_2006 ~~ health_2006
  depression_2008 ~~ health_2008
  depression_2010 ~~ health_2010
'

[Jeremy Miles]

You are regressing variables and correlating the same variables.

  depression_2002 ~ drug_2000 + health_2000 + depression_2000

And 

  depression_2002 ~~ depression_2000

In the same model will always lead to non identification.

When this happens, look closely at the output you get and see if the model tries to estimate parameters you aren't expecting (such as a correlation between depression at 2 waves).

Jeremy

--
You received this message because you are subscribed to the Google Groups "lavaan" group.
To unsubscribe from this group and stop receiving emails from it, send an email to lavaan+un...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/lavaan/3061fd0d-4450-4f08-97f0-079b8d79c029n%40googlegroups.com.