Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Wald test latent profile analysis

    Dear Stata users,
    this is just a confirmation question. I have run a latent profile model with 6 classes. l would like to conduct the Wald test to verify the equality of effects between classes. I am not sure if what I did is correct (it seems so, but I am not sure). As the command
    Code:
     estat ginvariant
    does not work, I made the following (as reported in page 480 of the manual https://www.stata.com/manuals/sem.pdf

    Code:
     gsem, coeflegend
    and then, considering for example the variable retired, I typed:

    Code:
     test _b[1.C: retired] = _b[2.C: retired] = _b[3.C: retired] = _b[4.C: retired]​​​​​​​ = _b[5.C: retired]​​​​​​​ = _b[6.C: retired]​​​​​​​
    I get the following:

    Code:
    [1b.C]o .retired - [2.C]retired = 0
    [1b.C]o .retired - [3.C]retired = 0
    [1b.C]o .retired - [4.C]retired = 0
    [1b.C]o .retired - [5.C]retired = 0
    [1b.C]o .retired - [6.C]retired = 0
    
    chi2 ( 5) = 84.86
    Prob > chi2 = 0.000
    Is the procedure correct? Can I conclude that the effect of the retired variable differs significantly across classes?
    ​​​​​​​Thank you very much
Working...
X