My MIMIC model is no identified

[Kiren CHEN]

I'm using sem to implement a mimic model in a research paper, which looks quite simple as follows:

However, i just make a toy model but it tells me that the model is not identified. It looks like this:

I appreciate that if someone can give any advice on this issue .

[Joseph Coveney]

1. In the research paper, it looks like "Environmental variables" is plural (i.e., there are many). You have one poi2.

2. I don't understand why you have error variances specified for the two latent factors. I would have made the correlation (covariance) between them as direct.

3. Be sure to check the paper to see whether they added some identifying constraints somewhere in order to accommodate all of those free covariances between the indicator variables,

You could start out with something like the following and see whether sem converges.

sem ///
    (poi1 poi3 poi5 <- X1) ///
    (poi2 you could stand more indicator variables here <- X2) ///
    (class4 <- X1 X2), covariance(X1*X2)

Also,

[Kiren CHEN]

Thanks Joseph.

The given code could work but seems to be different from the paper. In the paper, the arrow is from Environmental variables to Environmental dissimilarity, which looks to have a mimic model structure. Correspondingly, I think the arrow should from poi1 to X1, which may be

 
 sem ///     (poi1 poi3 poi5 -> X1) ///     (poi2 you could stand more indicator variables here -> X2) ///     (class4 <- X1 X2), covariance(X1*X2)

In this way, the code cannot work saying

'model not identified;
too many latent variables
r(503);
'
I have also checked the paper, there are no other constraints on this model.

Looking forward to your kind reply.

Thanks

[Joseph Coveney]

Yes, you're correct; my apologies. I was thrown because there is only a single variable on the right side that apparently is supposed to be the sole indicator for the two latent factors, and I misconstrued those on the left side as reflective without paying close enough attention to the direction of the arrows. The figure shown in the research paper doesn't make sense to me as a MIMIC model.