Sample weights with tidyLPA

[Jamie Tock]

Hello, 

Does anyone know of a method of entering sampling weights such as jackknife in conjunction with a tidyLPA model?  Thanks. 

Best,

Jamie Tock, PhD
Post-Doctoral Researcher
   Florida Center for Reading Research
   Florida State University
Affiliate Faculty
   South Central Mental Illness Research, Education, and Clinical Center
   New Orleans, LA

[Caspar van Lissa]

I'm not 100% clear on what you want to do, but it is possible to pass additional arguments to mclust() or mplusObject(), depending on which software you're using as estimator. 

[Jamie Tock]

I apologize.  I should have been more specific.  In Mplus I was able to use JACKKNIFE2 to include the sampling weights in my latent class analysis (see below).  I was hoping to do the same thing in tidyLPA.  From my understanding, some functions through mclust and mplusAutomation are possible with tidyLPA but not others.  I was hoping to be able to include sampling weights when estimating the latent profile analysis in tidyLPA. 

    WEIGHT = SPFWT0;

    REPWEIGHT = SPFWT1-SPFWT80;

     Missing = ALL (-999);

  USEOBSERVATIONS = C_D05 LT 3;

  categorical are C_D05 AgeBand;

 

    ANALYSIS:

      TYPE IS  COMPLEX;

      REPSE = JACKKNIFE2;

      Processors = 2;

 

      Model:

      Thinking by F_Q05a F_Q05b F_Q03a F_Q03c;

      Working by F_Q02b F_Q02c  F_Q02a F_Q04a;

      WorkTools by G_Q05a G_Q05c  G_Q05d G_Q05e  G_Q05f   G_Q05h;

      LifeTools by H_Q05a   H_Q05c  H_Q05d H_Q05e  H_Q05f  ;

[Jamie Tock]

Is it possible to add arguments to VARIABLE?  I have seen examples of adding arguments to ANALYSIS and OUTPUT but not VARIABLE.  I just want to add a weighting term like this:

  df3 %>% 

      estimate_profiles(5,

                        package = "mplusAutomation",

                        variances = c("equal","varying","equal","varying","equal","varying"),

                        covariances = c("zero","zero","equal","equal","varying","varying"),

                        select_vars = c("compsk","numsk","readsk","writsk"),

                        VARIABLE = "WEIGHT IS Weight0;")

But I am not sure if this is possible yet. 

On Monday, February 22, 2021 at 2:54:36 PM UTC-6 c.j.va...@gmail.com wrote:

[Caspar van Lissa]

This is not yet possible, because all columns in the data are used as indicators. I think the function MplusAutomation::createMixtures() might be more helpful; it provides less automation but more freedom.