Latent Class Growth Analysis using Lavaan

[Aron Lindberg]

Hi,

I'm trying to identify latent growth classes (similar to what Qureshi & Fang 2010 did). I understand from the manual that this should be possible, but don't quite know how to do it:

model <- ' 

c1(k=2) =~ t03 + t04 + t05 + t06

'

classes <- growth(model, data=latent.growth.data.formatted)

This returns:

Error in parse.rhs(rhs = lhs.formula[[2L]]) : 

  lavaan ERROR: I'm confused parsing this line: c12

Am I using the right syntax to generate a latent class growth model? How can I improve it so as to not get the error message above. Sorry for the vague question, but the manual does not give any clear guidance on latent class growth analysis. Maybe there are resources that I have missed?

Best,

Aron

P.S. If this does not work with lavaan, can anyone recommend a different package?

Reference

Qureshi, I., and Fang, Y. 2010. “Socialization in Open Source Software Projects: A Growth Mixture Modeling Approach,” Organizational Research Methods (14:1), pp. 208–238.

[Jan Lammertyn]

Dear Aron,

I'm not acquainted with growth curve models, but the "c1(k=2)" notation in your model looks suspicious to me.  Did you take a look at the tutorial http://users.ugent.be/~yrosseel/lavaan/lavaanIntroduction.pdf ?

Best,

Jan

[yrosseel]

On 01/18/2013 04:56 PM, Aron Lindberg wrote:
> Hi,
>
> I'm trying to identify latent growth classes (similar to what Qureshi &
> Fang 2010 did). I understand from the manual that this should be
> possible, but don't quite know how to do it:
>
> |
> model <- '
> c1(k=2) =~ t03 + t04 + t05 + t06
> '

Well, this syntax is certainly not described in the 'official' manual(s)
(on the lavaan.org website). I did once use the 'c1(k=2)' in a
presentation just as an example of how the syntax for discrete latent
variables could look like.

But to be clear: the current version of lavaan (0.5-11) does not yet
support discrete latent variables, and hence can not be use to estimate
latent class type models.

Yves.

[Aron Lindberg]

Thanks for letting me know. Yes, I took the syntax from a presentation of yours that I found...