Correlations between endogenous variables

[Marta Vigier]

Dear Lavaan  Experts,

I am a beginner to Lavaan and SEM and would like to very kindly ask for your help.

I tried to estimate a longitudinal Actor Partner Interdependence Model for one single variable (heart rate) for indistinguishable dyads. I am interested in partner effects but also in concurrent synchrony (correlation between two subjects within each time point). When I estimated my model (please see the figure) I got a correlation only within Time1 and residual covariances within Time2 and Time 3 (so only the partial correlations). When I estimated the same model but only for two time points (Time2 and Time3 and Time3 and Time1) I got signifiant correlations for all three time points. This would mean a concurrent synchrony in heart rate within time points. Is there any way to obtain correlations within each time points directly in my three time points model? If not, is it still possible to report the values that I found by treating each time point as exogenous variable?

In advance, thank you very much for your help.

Best regards,

Marta

[Terrence Jorgensen]

Correlations are not estimated parameters, except by coincidence (e.g., no predictors of Time 1 variables, so nothing is partialed out). Residual (partial) correlations are estimated parameters, controlling for any estimated effects.  So your 2-occasions models differ because you are partialing out different effects (either Time 1 or Time 2) when estimating the residual correlation at Time 3.  

If you just want to obtain the model-implied zero-order correlations, you can get it after estimation using

lavInspect(fit, "cor.ov")

Terrence D. Jorgensen

Assistant Professor, Methods and Statistics

Research Institute for Child Development and Education, the University of Amsterdam

http://www.uva.nl/profile/t.d.jorgensen