Adjusted R-Squared

[kretzsc...@gmail.com]

Dear all,

I would like to know if there is an aption to use an adjusted R-Squared (like in multiple regression) within lavaan. However, I'm even not sure if the adjusted R-Squared is common within the SEM approach; see the answer from Muthen a couple of years ago:
http://www.statmodel.com/discussion/messages/23/929.html?1134088447
I used Google to get more information but was not quite successful.

Thanks a lot and best wishes
André

[Yves Rosseel]

On 09/18/2014 10:38 AM, kretzsc...@gmail.com wrote:
> I would like to know if there is an aption to use an adjusted R-Squared
> (like in multiple regression) within lavaan.

I do not know of any literature on that (adj R^2 in SEM). I guess if you
have a recursive model with only observed variables, you could manually
compute the R^2 for each dependent variable, by using the usual formula:

adj(R^2) = 1 - (1-R^2) * (n - 1)/(n - p - 1)

where n is the total sample size, and p the number of predictors for
this dependent variable.

But in a general SEM model, it is not clear what 'p' is.

Yves.