identification SEM model Can someone explain to me how I calculate the non-redundant parameters? I read everywhere that you calculate this by p (p + 1) / 2, 
but do you have to do this per latent variable or in total or with aptitude and social support combined (since they covary)? 
Thank you!

 A: Welcome to CV, Michelle.
The formula you have supplied is used to calculate how many unique variances and covariances there are in your entire matrix of observed variables--it includes all variables in your model. These are the the "known" pieces of statistical information, if you will, for your model. The form of the model you specify has no impact on this initial matrix of observed variances/covariances; your modelling choices (e.g., latent covariance between aptitude and social support) are instead reflected in your model-implied variance/covariance matrix. We then later use these two matrixes (and the differences between them) to calculate indexes of model fit. 
You can calculate the degrees of freedom for your SEM by subtracting the number of parameters you wish to estimate--the "unknowns"--from your total pieces of "knowns". If your "unknowns" exceed your "knowns", your model will be unidentified, and you will not be able to estimate model fit or model parameters. 
In this case, you have 16 observed variables, and therefore 136 variances/covariances. You then are attempting to estimate:


*

*11 factor loadings (you have fixed one factor loading for each latent variable to a value of 1--I can't see which you fixed for performance in school, but I assume it is there)

*16 residual variances/error terms (one for each indicator)

*4 residual covariances (between indicators of happiness and previous happiness)

*5 latent variances/disturbances (I'm assuming five, but only two are depicted)

*1 latent covariance (between social support and aptitude)

*7 latent slopes
You are therefore estimating 44 parameters, which should produce a model with 92 degrees of freedom.
